university of kentucky uknowledge
TRANSCRIPT
University of Kentucky University of Kentucky
UKnowledge UKnowledge
DNP Projects College of Nursing
2015
Comparing MUST and the NRI Tools in the Identification of Comparing MUST and the NRI Tools in the Identification of
Malnutrition in Heart Failure Patients Malnutrition in Heart Failure Patients
Cassondra D. Degener University of Kentucky, [email protected]
Right click to open a feedback form in a new tab to let us know how this document benefits you. Right click to open a feedback form in a new tab to let us know how this document benefits you.
Recommended Citation Recommended Citation Degener, Cassondra D., "Comparing MUST and the NRI Tools in the Identification of Malnutrition in Heart Failure Patients" (2015). DNP Projects. 36. https://uknowledge.uky.edu/dnp_etds/36
This Practice Inquiry Project is brought to you for free and open access by the College of Nursing at UKnowledge. It has been accepted for inclusion in DNP Projects by an authorized administrator of UKnowledge. For more information, please contact [email protected].
FINAL DNP CAPSTONE REPORT
Comparing MUST and the NRI Tools in the Identification of Malnutrition in Heart
Failure Patients
Cassondra D. Degener, BSN, RN, RD
University of Kentucky
College of Nursing
May 2015
Melanie Hardin-Pierce, DNP, APRN, ACNP-BC Committee Chair/Academic Advisor
Fran Hardin-Fanning, PhD, RN Committee Member
Darlene Welsh, PhD, MSN, RN Committee Member/Clinical Mentor
Dedication
I would like to dedicate this final capstone project to my husband Kyle Degener. You
have been my inspiration, my rock and motivation for the last five years. Thank you for
encouraging me to pursue my dreams of becoming a Nurse Practitioner, and for all the
sacrifices you have made to help me accomplish this endeavor. Without you all this
would not be possible.
iii
Acknowledgements
I would like to acknowledge the following people for their assistance with various
aspects of this project:
Dr. Melanie Hardin-Pierce (academic advisor & committee chair): for serving as my
advisor over the past five years, for supporting me and guiding me through my
educational journey; for the time she spent reading and reviewing my work, and for her
dedication to educating future nurse practitioners.
Dr. Fran Hardin-Fanning (committee member): for her dedication to her students, for
her time spent reading, reviewing, and editing my manuscripts, and for her continued
support and encouragement.
Dr. Darlene Welsh (clinical mentor): for taking the time to read, review, and edit my
manuscripts and lend your expertise throughout my capstone processes
Amanda Green (core measures coordinator): for supplying me with the necessary
patient information for my project
Susan Westneat (statistician): for taking her time to assist me with data analysis for my
final project
Amanda Wiggins (statistician): for taking time to guide me through the statistical
analysis of my project
Whitney Kurtz-Ogilvie (writing specialist): for her time spent reviewing my work and
improving my academic writing skills
iv
Table of Contents
Acknowledgements……………………………………………………………….. iii
List of Tables…………………………………………………...…………………. v
Introduction to Final DNP Capstone Report…………………………………….... 1
Manuscript 1………………………………………………….…………………… 5
Manuscript 2………………………………………………………………….…… 38
Manuscript 3………………………………………………………………………. 63
Final DNP Capstone Report Conclusion………………………………………….. 94
Appendix A: Data Collection Tool………………………………………………... 97
References…………………………………………………………………………. 98
v
List of Tables
Manuscript 1
Table 1: Review of the MUST Literature …………………………………….……. 18
Manuscript 2
Table 1: Review of the NRI Literature .……………………………………………. 50
Manuscript 3
Table 1: Patient demographics ……………………………….…………………...... 84
Table 2: Blood biochemical measures ……………………………………………... 86
Table 3: Malnutrition prevalence …………………………………………………... 87
Table 4: Malnourished patient demographics …………………………………….... 88
Table 5: Malnourished patient lab and biochemical measures ..………………….... 90
Table 6: Prediction accuracy: Albumin, MUST and NRI compared to prealbumin…91
Table 7: Prediction accuracy: MUST compared to NRI ………………………….... 91
1
Introduction to Final DNP Capstone Report
Cassondra Degener
University of Kentucky
2
Heart failure (HF) is one of the top five leading causes of death in the United
States and each year roughly 825,000 people are newly diagnosed (Go et al., 2013). It is
estimated that one million Americans with decompensated HF are admitted to the
hospital every year, which contributes to over $35 billion in healthcare costs (Chaudhry
et al., 2013). The high incidence of hospitalizations and increased healthcare costs
among HF patients may be attributed to a number of causes, including malnutrition
(Lemon et al., 2009). Malnutrition prevalence in HF patients is as high as 66% (Aziz et
al., 2011).
The dietary intake and quality of those with HF is poor, which may lead to
damaging effects on disease progression and overall health status (Lemon et al., 2009;
Arcand et al., 2009). Poor dietary intake may be attributed to diminished appetite and
early satiety due to hepatic and gastrointestinal congestion, which is common in this
population (Kalantar-Zadeh, Anker, Horwich, & Fonarow, 2008; Nicol et al., 2002).
Other reasons for inadequate nutrient intake may be attributed to dietary restrictions,
fatigue, shortness of breath, nausea, anxiety and depressed mood (Lennie, Moser, Heo,
Chung, & Zambroski, 2006). Dietary intake in HF patients may be calorically the same
as healthy individuals, but they differ significantly in macro and micronutrient
composition (Machado d’Almeida, Perry, Clausell, & Souza, 2015). This lack of macro
and micronutrients can be detrimental to the overall health status of patients and lead to
worsening disease progression and outcomes (Machado d’Almeida et al., 2015).
Early identification of malnutrition is important to improving outcomes and
overall nutritional status of patients (Corkins et al., 2014). Traditional measures of
nutrition status such as laboratory (i.e. serum albumin and prealbumin) and
3
anthropometric measures (i.e. body mass index and percentage weight loss) are beneficial
in identifying malnutrition; however, they are not enough and can delay the recognition
of malnutrition, especially in HF patients (Araujo, Lourenco, Rocha-Gonocalves,
Ferreira, & Bettencourt, 2011; Corkins et al., 2014). To assist with the early
identification of malnutrition, researchers have developed a number of different screening
tools. Subjective screening tools can be rapid, easy and inexpensive ways to identify
malnutrition among hospitalized patients. With all the subjective screening tools
available, there are few studies available which evaluate the best methods of identifying
malnutrition in the HF population. Two screening tools, the Malnutrition Universal
Screening Tool (MUST) and Nutritional Risk Index (NRI), have shown some potential to
be reliable methods of evaluating the nutritional status of HF patients.
This practice inquiry project, through a retrospective electronic medical record
review, evaluated the presence of malnutrition in 100 HF patients admitted to the
University of Kentucky Chandler Medical Center. The primary goal of this project was
to test the performance of albumin, NRI and MUST in comparison to the reliable
screening measure of prealbumin, among HF patients admitted to the hospital. The
objectives were to (i) evaluate HF patients for the presence of malnutrition using four
screening measures (i.e., albumin, prealbumin, NRI and MUST), and (ii) determine
laboratory and co-morbidity trends among malnourished patients.
This evaluation project of HF patients will provide insight and guide further
research on effective objective and subjective screening methods that may help in the
identification of malnutrition in hospitalized patients with HF. This practice inquiry
4
project includes three manuscripts each of which discuss relevant aspects of malnutrition
and HF, and the best methods to screen for malnutrition.
Manuscript one is a literature review that examined the available studies in which
MUST was evaluated and compared to other, similar screening tools and objective
nutritional methods. Sixteen studies were evaluated in the review with respect to
MUST’s effectiveness in screening malnourished patients in multiple patient
populations in hospital and outpatient settings.
Manuscript two evaluated the available literature on NRI and provided evidence to
support whether or not it is reliable in various populations. The review evaluated ten
studies which compared NRI to other reliable screening tools and made
recommendations for practice.
Manuscript three evaluated hospitalized HF patients for the presence of malnutrition.
Four screening measures were used in order to determine laboratory and co-morbidity
trends among malnourished HF patients admitted to the University of Kentucky
Chandler Medical Center.
5
Manuscript 1
Malnutrition Universal Screening Tool: A Review of the Literature
Cassondra Degener RD, BSN, RN, CCRN, DNP Student
University of Kentucky
6
Abstract
Purpose: The purpose of this literature review was to find the studies available
evaluating the Malnutrition Universal Screening Tool (MUST) and comparing it to
similar screening tools and objective nutritional methods. The strengths, weaknesses and
reliability of MUST will be evaluated in comparison to other reliable screening tools, and
recommendations for practice will be provided.
Design and Methods: Literature review to find the available studies from 2004-2014
published in English using the databases of EBSCOhost, Academic Search Complete,
CINAHL, and MEDLINE. Ancestry searching was used to find additional articles
meeting the above criteria.
Results: Overall the search produced 52 articles, but only sixteen met the inclusion
criteria. Of the studies included in this review, six screened hospitalized patients, two
evaluated outpatients, seven articles examined chronic diseases (cancer, HF, and renal
failure), and one assessed elderly patients.
Practice Implications: The MUST has demonstrated evidence of reliability and validity
in multiple patient populations including outpatients, hospitalized, cancer, elderly, and
chronic disease. Many researchers noted the best nutritional screening methods were
those which combined a subjective screening tool and objective measures. There is a
growing need for studies that evaluate MUST and multiple subjective screening methods
against objective measures in the HF population.
Search Terms: Adult, elderly, cancer, chronic disease, heart failure, hospitalized
patients, malnutrition, malnutrition universal screening tool, screening tools, and
surgery.
7
Background and Significance
Malnutrition is a major health problem in the United States. The prevalence of
malnutrition is 23% among hospitalized inpatients, with malnourished patients spending
4.5 days longer in the hospital compared to well-nourished ones (Gout, Barker, & Crowe,
2009). The increased hospital length of stay can triple healthcare costs from $9,485 for
the average hospitalized patient to $26,944 for malnourished ones (Corkins et al., 2014).
Disease-related malnutrition occurs with chronic diseases such as rheumatoid arthritis,
cancer, renal failure and HF (Jensen et al., 2010).
Heart failure is one of the top five leading causes of death in the United States
(Go et al., 2013). Every year roughly 825,000 people are newly diagnosed with HF and
one in five will die within one year of diagnosis (Go et al., 2013). Heart failure accounts
for 1 million hospital admissions per year, with an average length of stay (LOS) of 4.9
days (Centers for Disease Control, 2013).
Malnutrition is highly prevalent in the HF population and can lead to a condition
called cardiac cachexia (Hoes, 2007). Roughly 15 percent of HF patients will develop
cardiac cachexia, which is associated with a poor prognosis (Hoes, 2007). Cardiac
cachexia is responsible for increased morbidity and mortality, and decreased quality of
life among patients with HF (Moughrabi & Evangelista, 2007). There is a growing need
for a reliable, easy to use screening tool that can be used in the HF population which will
assist health providers to identify and treat malnourished patients.
Many instruments are available to help evaluate nutrition risk in hospitalized
patients. The Malnutrition Universal Screening Tool, or MUST, was originally
developed by the Malnutrition Advisory Group for the British Association of Parenteral
8
and Enteral Nutrition (Elia, 2010). The MUST is a five-step tool that evaluates BMI
score, recent weight loss and acute disease, assigns an overall numerical risk, and then
provides management guidelines (Elia, 2010). Step one is the BMI category which
provides scores as follows: BMI >20 = 0 points, 18.5 – 20 = 1 point, and <18.5 = 2 points
(Appendix A). Step two provides a weight loss score based on the amount of weight lost
in the past 3-6 months: a score of 0 for 5%, 1 for 5 – 10%, and a 2 for >10%. Step three
is to determine if the patient has been acutely ill and if there has been or is likely to be no
nutritional intake for >5 days which provides a score of 2. For step four, the user adds all
point scores together: a total score of 0 = low risk, 1 = medium risk, and 2 or more = high
risk. Step five provides appropriate management guidelines based on the overall
malnutrition risk score. Patients who score a 0 are at a low nutritional risk, and no
interventions are necessary. A score of 1 indicates moderate risk patients and close
dietary monitoring is recommended. A score of 2 or more means the patient should have
a complete nutrition assessment by a registered dietitian (Elia, 2010). One benefit of the
MUST tool is that it guides the user to either seek immediate nutrition consultation for
high risk patients, or to observe medium risk patients upon hospital admission.
Purpose of the Integrative Review
The purpose of this literature review was to examine the available studies in
which MUST was evaluated and compared to other, similar screening tools and objective
nutritional methods (i.e. albumin and prealbumin). This paper also seeks out to
determine if MUST is reliable in screening for malnutrition in multiple patient
populations including HF. Sixteen studies were evaluated in the following review with
respect to MUST’s effectiveness in screening malnourished patients in multiple patient
9
populations in hospital and outpatient settings. The strengths, weaknesses and reliability
of MUST were evaluated in comparison to other reliable screening tools and practice
recommendations were made. MUST was chosen over other screening measures because
of its ease of use and rapid completion by the user, making it practical to use in any
healthcare setting.
Methods
Search Method
The EBSCOhost, Academic Search Complete, CINAHL, and MEDLINE
databases were searched through the UK Libraries website. Inclusion criteria involved
published studies which compared MUST to other reliable screening methods in adult
patients in multiple settings such as the hospital, outpatient clinics, or long term care
facilities. The search only included articles from 2004-2014 which were either published
in English or translated into English. Keywords used in the search included adult,
elderly, cancer, chronic disease, heart failure, hospitalized patients, malnutrition,
malnutrition universal screening tool, screening tools, and surgery. I used ancestry
searching to find additional articles meeting the above criteria. Studies were excluded if
they were published before 2004, not written or published in English, and if they did not
compare MUST to other nutritional screening methods.
Search Outcome
Overall the search produced 52 articles, but only sixteen met the inclusion criteria.
Of the studies included in this review, six studies involved hospitalized patients, two
evaluated outpatients, and one screened the elderly. Disease specific studies included
four oncology, two cardiac and one renal. The cardiac studies consisted of patients
10
undergoing heart surgery, not specific medically managed HF patients. Of the sixteen
studies, eight were prospective, five were cross-sectional, two were observational, and
one was longitudinal.
Findings and Synthesis of Themes
Several themes emerged during this literature review with regards to screening
tools and practice recommendations for nutritional screening. In addition to MUST, a
number of subjective screening tools were compared in the studies including the
following: Subjective Global Assessment (SGA), Patient-Generated SGA (PG-SGA),
Nutrition Risk Screening 2002 (NRS-2002), Nutritional Risk Index (NRI), Geriatric NRI
(GNRI), Malnutrition Screening Tool (MST), Modified MST (Mod-MST), Short
Nutritional Assessment Questionnaire (SNAQ), Mini-Nutritional Assessment (MNA),
MNA Short Form (MNA-SF), and the Cardiac Surgery Specific MUST (CSSM). Overall
ten studies recommended specific screening tools for use in malnutrition risk assessment,
while two suggested anthropometric or objective measures, two proposed a combination
of subjective screening tools and objective measures, and two recommended further
research as opposed to any specific screening measures. The following section will
illustrate those themes. Refer to Table 1 for specifics about each study and their
limitations.
MUST Recommended Alone or in Combination with Other Screening Tools
Of the sixteen studies in this review, seven found MUST to be reliable. Four of
those seven recommended MUST alone, while three suggested MUST and other
subjective screening tools. Stratton et al. (2004) evaluated eight different screening tools
among medical and surgical patients and found MUST and MST to be the easiest tools to
11
complete, according to feedback from nurses, nutritionists and medical students. Based
on their statistical analysis (Table 1), overall MUST showed high validity (Table 1)
compared to the other evidence based malnutrition diagnostic tools (Stratton et al., 2004).
Poulia et al. (2012) on the other hand conducted their study to evaluate the nutritional
status of hospitalized elderly patients using six screening tools. The authors found
MUST and MNA-SF to be the most reliable with sensitivities of 87.3% and 98.1%
respectively, but MUST and SGA demonstrated the best agreement compared to the gold
standard, with kappa values of 0.64 and 0.71 respectively (Poulia et al., 2012). Both
studies suggested the use of MUST over all the other tools used in their studies (Stratton
et al., 2004; Poulia et al., 2012).
Tu, Chien, and Chou (2012) compared MUST, NRI and SGA in their study
comprised of forty five colorectal cancer patients. They demonstrated MUST and NRI to
be comparable measures with higher sensitivities (96.0% & 95.2%) and lower
specificities (75% & 62.5%). The authors found MUST to be easier to complete,
inexpensive and faster compared to NRI and SGA (Tu et al., 2012). Another study also
compared three screening tools in the oncology setting (Amaral, Antunes, Cabral, and
Kent-Smith, 2008). The MUST showed the highest agreement with the reference tool of
NRS-2002 based on its sensitivity of 97.3%, specificity of 77.4% and kappa agreement of
0.64 (Amaral et al., 2008). The authors also recommended MUST based on its reliability
(Amaral et al., 2008; & Tu et al., 2012).
Among hospitalized patients, Velasco et al. (2010) and Kyle, Kossovsky,
Karsegard, and Pichard (2006) compared three different screening tools using SGA as the
standard. Velasco et al. (2010) found good agreement between NRS-2002 and SGA
12
(kappa 0.62) and MUST and SGA (kappa 0.64). Based on their statistical analysis (Table
1), both studies demonstrated that MUST and NRS-2002 were the most reliable tools and
one or the other should be implemented for malnutrition screening upon hospital
admission. In addition to MUST and NRS-2002, one study also recommended the use of
SGA (Kyle et al., 2006).
Vicente et al. (2013) evaluated gastric and colorectal patients using NRI, MUST,
MST, SGA, BMI and albumin. Statistical analysis showed MUST had a sensitivity and
specificity of 84% and 73.4% respectively. Vicente et al. (2013) suggested MUST and
SGA were the best screening methods among cancer patients. These seven studies
demonstrated MUST to be reliable in multiple populations including medical, surgical,
cancer and elderly patients. The authors recommended MUST alone or in combination
with NRS-2002 or SGA.
Alternative Screening Tools Recommended for Practice
Neelemaat et al. (2011) evaluated hospitalized medical and surgical patients to
compare six subjective tools and two anthropometric measures. They found MST and
SNAQ to be faster and easier tools, when compared to the more comprehensive tools of
MUST and NRS-2002. The MST and SNAQ had adequate sensitivities and specificities
of ≥70%, but their scores were slightly lower than the other tools. Based on all the
available information, the authors suggested the use of either MST or SNAQ upon
hospital admission (Neelemaat et al., 2011). Like Neelemaat et al. (2011), hospitalized
medical and surgical patients were also screened by Olivares et al. (2014), but the authors
only used four subjective tools. The authors found the NRS-2002 and MNA-SF to be
highly reliable measures compared to SGA, with kappa values of 0.57 and 0.67
13
respectively (Table 1; Olivares et al., 2014). In contrast to other studies, the authors
suggested the use of NRS 2002 because it was the easiest and took the least amount of
time to complete (Olivares et al., 2014).
One study evaluated the nutritional status of hospitalized medical patients using
three screening tools (Gibson, Sequeira, Cant and Ku, 2012). Based on statistical
analysis, MUST and Mod-MST had sensitivities of 80% and 77% respectively, with
specificities of 85% and 83% (Gibson et al., 2012). Both tools had 29 false negatives, but
MUST had 14 false positives while Mod-MST had 16. Noting similar scores between the
tools, the authors suggested the use of the Mod-MST on hospital admission given it was
easier and faster to use (Gibson et al., 2012). In these three studies evaluating surgical
and/or medical patients, the authors recommended four different subjective tools
including MST, SNAQ, NRS-2002 and Mod-MST (Neelemaat et al., 2011; Olivares et
al., 2014; & Gibson et al., 2012).
Alternative Objective Measures Recommended for Practice
One study did not recommend MUST, but rather objective screening measures
(Boleo-Tome, Monteiro-Grillo, Camillo, & Ravasco, 2011). The authors evaluated the
nutritional status of cancer patients using objective measures and two subjective tools.
Results indicated that MUST was the best tool for routine screening in radiation cancer
patients given its sensitivity of 80% and specificity of 89% (Boleo-Tome et al., 2011).
Given the time constraints of health professional however, the authors suggested the use
of percent weight loss over the last 3-6 months to be used with admission screenings
(Boleo-Tome et al., 2011).
14
In contrast to Boleo-Tome et al. (2011), Leistra et al. (2013) compared objective
and subjective methods in the outpatient setting and did not find MUST or the other
subjective tool (SNAQ) to be reliable. They found both subjective screening tools to
have insufficient validity, noting SNAQ identified too few and MUST too many patients
as being malnourished (Table 1). Their recommendation was to use anthropometric
measures and weight loss to better identify malnourished patients in the outpatient setting
(Leistra et al., 2013). These two studies suggest that subjective tools may be useful but
are not comparable to objective measures in terms of efficiency and reliability.
Combination of Screening Tools and Objective Measures
A few researchers noted the best nutritional screening methods were those which
combined a subjective screening tool and objective measures such as lab values and
anthropometric measures (Almeida, Correia, Camilo, and Ravasco, 2011; Van Venrooij
et al., 2011). Almeida et al. (2011) compared four screening tools to the objective
measures of percentage weight loss and BMI in surgical patients to determine their
nutritional status. Van Venrooij et al (2011) on the other hand, used four screening tools
and two objective measures to screen cardiac surgery patients. Based on statistical
analysis, found the NRS-2002 and MUST tools to be the most concordant with
sensitivities of 80% and 85% respectively (Table 1; Almeida et al., 2011). The authors
went on to suggest the combination of either NRS-2002 or MUST with percentage
weight loss on admission (Almeida et al., 2011). In contrast, Van Venrooij et al. (2011)
found the CSSM tool to be the most reliable, noting a sensitivity and specificity of 74.1%
and 70.1% respectively. They too recommended a combination of objective and
15
subjective screening measures in the cardiac surgery population in order to accurately
identify those who may truly be malnourished.
No Specific Tool Recommendations
In two studies screening chronic disease patients, the authors did not find
sufficient evidence to recommend a specific subjective screening tool. Lawson et al.
(2012) evaluated the nutritional status of renal patients using the three screening tools,
while Lomivorotov et al. (2013) compared SGA to three screening tools in cardiac
surgery patients. Lawson et al. (2012) found MUST and MST were not sensitive enough
for all types of renal patients with sensitivities of 53.8% and 48.7% respectively. They
did note both tools showed fair reliability compared to anthropometric nutritional
markers (Table 1). Based on statistical analysis the authors did not recommend one
specific tool for nutritional screening, but did stress the need for larger studies which
evaluate multiple screening methods in renal patients (Lawson et al., 2012).
Lomivorotov et al. (2013) found SNAQ and MUST had comparable accuracy in
detecting malnutrition but not in predicting post-operative outcomes (Table 1). This led
the authors to not recommend a specific screening tool but suggest that more research is
needed to understand the use of nutrition screening tools in the HF and cardiac surgery
populations (Lomivorotov et al., 2013). The above research studies indicated more
research is needed to evaluate subjective screening tools among patients with chronic
diseases such as HF and renal failure (Lawson et al., 2012; Lomivorotov et al. 2013).
Practice Implications
Malnutrition is highly prevalent in hospitalized, chronic disease patient
populations but often remains unidentified and untreated (Lawson et al., 2012). The
16
overwhelming cost of malnutrition suggests the need for a consistent and reliable
nutrition screening method that is easy to use and transferable across multiple patient
populations (Elia, 2009). The Malnutrition Universal Screening Tool has demonstrated
evidence of reliability and validity in multiple patient populations including outpatients,
hospitalized, cancer, elderly, and chronic disease.
There appear to be major gaps in the literature involving a consistent and reliable
screening tool which can be used for patients with chronic diseases such as renal and HF.
The MUST was utilized in cardiac surgery patients, but not specifically in medically
managed HF patients. Through this literature review, MUST showed reliability and
validity in various patient populations. Implementing its use upon hospital admission
may help identify those at malnutrition risk earlier and possibly improve patient
outcomes. There is a growing need for studies that evaluate MUST and multiple
subjective screening methods in combination with objective measures (i.e. albumin,
prealbumin and recent weight loss) in the HF population.
Conclusions
No one tool has demonstrated consistent reliability and validity in screening for
malnutrition among multiple patient populations in various healthcare settings.
Malnutrition can occur in many patient populations including chronic diseases such as
cancer, liver failure and HF (Jensen et al., 2010). In HF, malnutrition can be as prevalent
as 36% based on serum albumin levels and the presence of less than 90% ideal body
weight (Nicol, Carroll, Homeyer, & Zamagni, 2002). The use of MUST in the HF patient
population is not well described in the literature; however, it has been used with success
17
in other adult and elderly populations. More research needs to be conducted within the
HF population to better identify a reliable and valid tool for this population.
18
Table 1 Review of the MUST Literature
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Almeida,
Correia,
Camilo,
&
Ravasco:
2011
Prospective
cross-
sectional
study, over
eight months
with all data
collected by a
single
research
dietitian to
determine
nutritional
status.
Three hundred
surgical
hospitalized
patients; ages
43 - 77; 44%
male; 46%
cancer patients
BMI, recent %
weight loss,
Nutrition Risk
Screening 2002
(NRS 2002),
Malnutrition
Universal
Screening Tool
(MUST),
Nutritional Risk
Index (NRI),
Subjective
Global
Assessment
(SGA)
Compared to
SGA the
Sensitivity,
Specificity,
PPV, NPV:
NRS 2002 -
80%, 89%,
87%, 100%;
MUST - 85%,
93%, 89%,
99%; NRI -
29%, 27%,
24%, 27%;
BMI - 43%,
39%, 35%,
31%; % wt loss
- 89%, 93%,
81%, 89%
NRS 2002 and
MUST are the most
concordant, valid
and reliable tools to
detect nutrition risk
in surgical patients.
>5% weight loss
over six months
was reliable and
valid. Percent
weight loss
estimation should
be mandatory in
routine practice to
increase outcome
driven nutrition
management.
MUST and NRI
were made into
two categories for
the purpose of the
study, but each
were originally
developed into
three and four
categories. Made
two categories in
order to determine
comparisons, but
only two
categories could
affect the results.
19
Table 1 (continued)
Author
and Year
Study Design and
Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Amaral,
Antunes,
Cabral,
Alvest, &
Kent-
Smith:
2008
Prospective study
over two months
at a
comprehensive
cancer center in
Portugal. One
researcher
collected all data
to determine
nutritional status
and the tools'
ability to predict
length of stay.
One hundred
thirty cancer
patients (head
and neck, GI,
GU, breast,
lymph,
endocrine,
respiratory,
bone); ages 43-
71; 44% female
MUST,
Malnutrition
Screening
Tool (MST)
& NRS 2002
Compared
to NRS
2002;
sensitivity,
specificity,
PPV, NPV,
kappa
agreement:
MUST -
97.3%,
77.4%,
63.2%,
98.6%, 0.64;
MST -
48.7%,
94.6%,
78.3%,
82.2%, 0.49
MUST is most
concurring with
NRS 2002 in
hospitalized cancer
patients and better
at identifying
patients at risk for
increased length of
stay. MST was a
better diagnostic
value in head/neck,
peritoneal and GI
cancers. The three
screening tools
agreed with respect
to identifying
head/neck cancer
patients at highest
nutritional risk.
MST was easiest
for patients to use
for self-screening
because it required
no training
compared to MUST
and NRS-2002.
Inherent to design,
patients admitted to
the study may not
represent the full
spectrum of cancer
patients. Small
sample size in some
diagnostic groups
compromised the
influence for some
types of patients.
Excluded critically
ill patients because
their nutritional
status would
seriously be
affected. But this
limited the
usefulness of the
studied tools in
such diagnoses.
20
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Boleo-
Tome,
Monteiro-
Grillo,
Camilo, &
Ravasco:
2011
Prospective
cross-sectional
study over ten
months to
classify
nutritional risk
and status
categories;
compare results
between
nutritional
parameters;
and validate
MUST in the
cancer
population. All
data collected
by single
research
dietitian.
Four hundred
fifty adult
cancer patients;
ages 18-95;
60% male;
predominantly
with breast,
prostate, lung
and colorectal
cancer
BMI, %
weight loss,
Patient
Generated-
Subjective
Global
Assessment
(PG-SGA),
and MUST
Compared to
SGA: MUST
sensitivity 80%,
specificity 89%,
PPV 87%, NPV
100%, kappa
0.86;
Percentage of
malnourished
patients: BMI
4%, SGA 29%,
and MUST
31%
MUST was
strongly
recommended to be
integrated in
routine screening in
the radiation
oncology setting.
It should be the
primary tool to
refer patients for
exact nutritional
screening with the
PG-SGA tool.
Based on time
constraints of
health
professionals, it is
recommended to
use % weight loss
in last 3-6 months
as a valid and
minimum
parameter to
predict nutritional
risk.
Included a
heterogeneous
population of
cancer patients in
terms of primary
site, nutritional
goals, radiologic
fields and
prognosis. Study
population was
restricted to
radiotherapy
patients and cannot
be generalized to
all cancer patients;
however it is a
good basis for
future studies in
oncology.
21
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Gibson,
Sequeira,
Cant, &
Ku: 2012
Prospective
study to
explore the
ease of use of
two screening
tools.
Compare the
validity in
adult acute
hospital
patients over 2
months in 2
separate
screening
phases.
Two hundred
sixty two
medical ward
patients; mean
age 70.8 ±
16.3yrs; 51.5%
female
MUST,
Modified
Malnutrition
Screening
Tool (Mod-
MST), SGA
Malnutrition risk
by tool: MUST
32.4%; Mod-MST
32.8%, SGA
26.7%. Compared
to SGA, sensitivity
& specificity:
MUST 80%, 85%;
Mod-MST 77%,
83%. False
negatives/false
positives: MUST
14/29; Mod-MST
16/29.
MUST and Mod-
MST were valid and
feasible to use with
medical patients.
Little variation
between the two
tools compared to
SGA, but sensitivity
and specificity were
85%. MUST took
up to five minutes
longer and Mod-
MST was easier to
use. Mod-MST was
recommended
because one needs
to choose tools that
are effective and
easy to use in
massive-screening
programs.
Large sample may
have more
confidently
predicted the two
groups of
misclassified
patients. There
were a number of
patients admitted
to the hospital but
missed in the
screening process.
Only three staff
members
completed the
interviews and
may need a larger
sample in order to
apply to other
groups.
22
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Kyle,
Kossovsky,
Karsegard
& Pichard:
2006
Population
study to test
the sensitivity
and
specificity of
three
screening
tools
compared to
SGA, to
assess the
association
between
nutritional
risk and
hospital LOS
over a 3
month period.
Nine hundred
ninety five adult
medical and
surgical
patients; 53%
male; mean age
50.5 21.9
(<10d LOS),
65.4 18.7
(>11d LOS)
NRI,
MUST,
NRS-
2002,
SGA
Moderate/severe
nutritional risk:
SGA 39%, NRI
25%, NRS-2002
28%, MUST 37%.
Compared to SGA,
sensitivity,
specificity, PPV,
NPV, kappa: NRI
43.1%, 89.3%,
76.2%, 66.3%, 0.24;
MUST 61.2%,
78.6%, 64.6%,
76.1%, 0.26; NRS-
2002 62%, 93.1%,
85.1%, 79.4%, 0.48
Significant
association
between LOS and
moderate to severe
malnutritional
status among all
tools. NRS-2002
had higher
sensitivity and
specificity
compared to SGA
than NRI and
MUST. The
authors
recommended
using the NRS-
2002, MUST and
SGA tools on
admission to
screen patients for
malnutrition.
SGA does not
allow for
categorization of
mild malnutrition
and focuses on
chronic not acute
malnutrition.
Screeners should
have been better
trained on the
screening tools
before
implementing the
study. LOS was
studied as an
outcome
parameter, but
many other factors
influence LOS, not
just malnutrition,
which were not
assessed in this
study.
23
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Lawson,
Campbell,
Dimakopoulos,
& Dockrell:
2012
Cross-sectional
and longitudinal
study to
determine the
validity and
reliability of two
screening tools
in renal patients
over six months
in a London
tertiary hospital.
Study divided
into three study
arms: 1)
concurrent and
predictive
validity; 2)
construct
validity; 3)
reliability.
Two hundred
seventy six
patients; in three
study arms: 1) 190
pts, median age
65, 48% female; 2)
46 pts, median age
61, 49% female; 3)
40 pts, median age
64, 48.5% female.
All patients
received peritoneal
or hemodialysis,
renal replacement
therapy, or
transplant.
SGA,
MUST,
MST
1) Validity
compared to
SGA, sensitivity,
specificity, PPV,
NPV, & k:
MUST - 53.8%,
78.3%, 73.7%,
60%, 0.316;
MST - 48.7%,
85.5%, 78.7%,
60.2%, 0.335. 2)
Risk of
malnutrition
classification:
MUST 22.5%;
MST 27.5%. 3)
Agreement
between repeat
tests, k value:
MUST 0.58
(moderate);
MST 0.33 (fair).
MUST and MST
not sensitive
enough to
identify all of the
malnourished
renal in-patients;
despite being
fairly reliable and
related to other
markers of
nutritional status.
There is a
growing need for
more research on
a renal-specific
nutrition
screening tool.
Need a larger
sample to
better
determine
reliability in
all renal
patients. Fluid
status could
not be
determined in
this patient
population
which may
skew patient
weights and
effect
nutritional
status
estimates.
24
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Leistra,
Langius,
Evers, van
Bokhorst-
de van der
Schueren,
Visser, de
Vet, &
Kruizenga:
2013
Cross-sectional
multicenter
study at nine
hospitals in the
Netherlands to
determine the
validity of
screening tools
in identifying
severely
undernourished
patients.
Two thousand
two hundred
thirty six
hospital
outpatients
patients; ages
40-72 years;
52.4% female
BMI, %
weight loss,
MUST, Short
Nutritional
Assessment
Questionnaire
(SNAQ)
BMI and %
weight loss - 6%
severe, 7%
moderate; MUST -
9% severe, 6%
moderate; SNAQ -
3% severe, 2%
moderate.
Sensitivity,
Specificity, PPV,
NPV
(severe/moderate):
MUST - 75%/82%,
94%/95%,
43%/71%,
98%/97%; SNAQ -
43%/29%,
99%/98%,
78%/72%,
96%/90%.
Validity of MUST
and SNAQ is
insufficient for
hospital
outpatients. SNAQ
identified too few
undernourished
patients, while
MUST identified
too many. It is
recommended to
measure weight,
height and weight
loss to better
determine
undernourishment
in hospital
outpatients.
Patients
completed the
assessment
forms, rather
than trained
medical
professionals.
There remains
an absence of a
gold standard
screening tool
with which to
compare other
tools. Only
two tools were
used in the
study and with
the variety of
tools available,
more could
have been
used.
25
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Lomivorotov,
Efremov,
Boboshko,
Nikolaev,
Vedernikov,
Lomivorotov,
& Karaskov:
2012
Prospective
cohort study
over eight
months to
assess the
prognostic
value of
different
screening tools
in
cardiopulmona
ry bypass
patients.
Eight hundred
ninety four
patients; > 53
years of age;
21% > 65
years; 37%
female, 14%
with Diabetes;
2.4% with EF
< 35%; 8.7%
redo surgery
SGA, NRS
2002,
MUST,
Mini-
Nutritional
Assessment
(MNA),
SNAQ
Compared to
SGA,
malnourished
patients,
sensitivity,
specificity,
PPV, NPV:
NRS 2002 -
6%, 38.3%,
95.4%, 31.6%,
96.5%; MUST
- 17%, 97.9%,
87.1%, 29.7%,
99.9%, MNA -
23%, 81.8%,
80.7%, 20.4%,
98.6%; SNAQ
- 17%, 91.5%,
87.5%, 28.9%,
99.5%;
SNAQ and
MUST have
comparable
accuracy in
detecting
malnourished
patients. MUST
independently
predict post-op
complications.
All tools were
insufficiently
sensitive to the
risk for
development of
post-op
complications.
Need to study if
pre-op nutrition
interventions will
improve patient
outcomes. Need
to develop more
sensitive methods
for screening this
population.
SGA is limited in
cardiac disease
because it relies on
the interviewer's
training and on the
interpretation of the
results, making it
less able to
reproduce in daily
clinical practice. It
has also been known
to miss acute
changes in
nutritional status and
miss some cases of
malnutrition. The
precise analysis of
body composition
using bioelectrical
impedance was not
performed, and can
affect the lack of
correspondence
between nutritional
screening results and
BMI. Long term
data was not
analyzed.
26
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Neelemaat,
Meijers,
Kruizenga,
van
Ballegooijen,
& van
Bokhorst-de
Vander
Schueren:
2011
Cross
sectional
screening to
compare five
malnutrition
screening
tools against
a reliable
screening
method in
one hospital.
Two hundred
seventy five
patients; 62%
over 60 yrs;
37% female
MST, SNAQ,
MNA short
form (MNA-
SF), MUST,
NRS 2002,
BMI,
unintentional
weight loss
No risk
compared to at
risk patients;
sensitivity,
specificity,
PPV, NPV:
MST 78%,
96%, 89%,
91%; SNAQ
75%, 84%,
66%, 90%;
MUST 96%,
80%, 69%,
98%; NRS
2002 92%,
85%, 72%,
96%; MNA-SF
100%, 41%,
42%, 100%.
The authors
suggested a
sensitivity and
specificity of
≥70% for a tool
to be considered
adequate.
MST and SNAQ
are quick and easy
tools and suitable
for use in hospital
inpatient settings
with sensitivity and
specificity 70%.
MST and SNAQ
performed well
compared to the
more
comprehensive
tools, MUST and
NRS 2002, on
criterion validity.
MNA-SF showed
great sensitivity but
low specificity in
the elderly
population. MUST
was less applicable
in the study because
there were a lot of
missing values.
Pre-set
definition of
malnutrition
(BMI and
weight loss)
could not be
determined in all
patients. Data
was completed
by trained
Dietitians, but
25% did not
have their
nutritional status
determined.
Selection bias
was excluded
because of this,
and the actual
rate of
malnutrition
could be higher.
27
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Olivares,
Ayala, Salas-
Salvado,
Muniz,
Gamundi,
Martinez-
Indart, &
Masmiquel:
2014
Prospective
study to
determine the
prevalence of
malnutrition,
identify
malnutrition risk
factors, and
compare
validity of tools
to the SGA in
hospitalized
patients during a
four month
period.
Five hundred
thirty seven
adult patients;
45% medical,
55% surgical;
ages 43-78;
56.4% male
SGA,
MNA-SF,
NRS
2002,
MUST
Compared to
SGA,
sensitivity,
specificity,
PPV, NPV, k-
value: MNA-
SF - 69.9%,
94.7%, 93%,
75.8%, 0.67;
NRS 2002 -
68.9%, 90.1%,
92.4%, 62.3%,
0.57; MUST -
64.1%, 91.9%,
91.5%, 65.3%,
0.56
Any of the tests
would be good to
use on admission
to screen for
malnutrition. NRS
2002 was chosen
because it was the
easiest and took the
least amount of
time. Difference in
malnutrition rates
between tools can
be explained by
severity of
underlying disease,
population setting,
and age. NRS-
2002, MNA-SF
and SGA have high
reliability. MUST
is invalidated after
adjusting for risk
factors because
weight loss and
low BMI are not
frequent in the
study population.
Could not be
extrapolated to other
hospitals in different
countries because it
was conducted in a
second level
hospital in Spain.
Could not assess
other population
types for
malnourishment
such as surgery or
transplant patients.
28
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Poulia,
Yannakoulia,
Karageorgou,
Gamaletsou,
Panagiotakos,
Sipsas, &
Zampelas:
2012
Prospective
study to
evaluate of
the efficacy
tools to
predict
malnutrition
in elderly
patients
admitted to
the hospital in
Athens,
Greece over
nine months.
Two hundred
forty eight
elderly patients
> 60 years;
mean age 75.2
+/- 8.5yrs; 52%
male; admitted
for neurologic
syndrome,
fever, blood
disease,
rheumatologic
disease,
malignancy,
hemorrhage,
diabetes, GI,
kidney or
respiratory
disease
NRI,
GNRI,
SGA,
MUST,
MNA-
SF, and
NRS-
2002
Compared to
true nutritional
status
(combined
index),
sensitivity,
specificity, PPV,
NPV, kappa:
NRI - 71.7%,
48.8%, 85.4%,
29.3%, 0.550;
GNRI - 66%,
92.1%, 94.6%,
56.45%, 0.465;
SGA - 84.3%,
91.4%, 95.2%,
74.3%, 0.707;
MUST - 87.3%,
76.8%, 88.4%,
75%, 0.638;
MNA-SF -
98.1%, 50%,
79.9%, 93.2%,
0.545; NRS
2002 - 99.4%,
6.1%, 68.2%,
83.3%, 0.088
The NRI was
higher in
sensitivity and
PPV than the
other tools, but
scored lower in
specificity and
NPV. MUST
and MNA-SF
were the most
valid. MUST
and SGA
showed better
agreement with
the combined
index. The
researchers
concluded that
the combination
of objective and
subjective
diagnostic tools
that are easy to
use are the best
for nutritional
screening.
Some patients had to
have the
questionnaires
translated for them and
results then had to be
translated again,
making for some
miscommunication
among patients and
researchers. Some
patients estimated
height and weight
instead of being
accurately measured
by researchers and in
4.8% of patients,
anthropometric
measurements were
not available. These
variations in accuracy
of measurements could
affect calculations and
results.
29
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Stratton,
Hackston,
Longmore,
Dixon,
Price,
Stroud,
King, &
Elia: 2004
Series of
prospective
studies (one
outpatient, four
inpatient
settings) to
assess the
prevalence of
malnutrition
risk between
MUST and
other screening
tools for
inpatients and
outpatients,
determine
concurrent
validity of
MUST and
other tools, and
the ease of use
of the screening
tools.
Among all series
of studies: a) 50
outpatients, b)
75 medical
inpatients, c) 85
surgical
inpatients, d) 86
elderly patients,
e) 50 medical
inpatients, and f)
52 surgical
inpatients.
Specific patient
demographics
not reported
MUST,
MEREC
Bulletin tool
(MEREC),
Hickson and
Hill tool
(HH), NRS,
MST, MNA-
SF, SGA, and
undernutrition
risk score
(URS)
Concurrent validity
= percentage of
patients placed in
same nutrition risk
category as MUST:
a) MEREC 92%,
HH 84%; b)NRS
89% (<65yrs), 92%
(>65yrs),MST 88%
(>65yrs); c) MNA
80%; d) MNA
77%; e) SGA
72%/92%; f) URS
67%/77%. Ease of
use of tools/time to
complete: MUST -
very easy and easy
(3-5min); MST -
very easy (3 min);
MNA - easy (5
min); NRS, HH (5-
7 min), SGA &
URS (5-10 min) -
difficult.
A desirable
screening tool
should be
rapid and easy
to use.
Results
indicate
MUST was
rapid and
easy/very easy
to use and
showed 'fair-
good' to
'excellent'
concurrent
validity with
most of the
other tools.
Bias in
concurrent
validity is
possible. The
sample
demographics of
the five separate
investigations
were not
disclosed. The
only thing known
about those
patients are the
age
classifications
and patient
cohorts (medical
or surgical,
inpatient or
outpatient). This
leaves the
inability to
reproduce the
same studies.
30
Table 1 (continued)
Author
and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Tu,
Chien,
&
Chou:
2012
Prospective
study to assess
the nutritional
status of
patients with
colorectal
cancer before
and after
surgery in
Taiwan over
two years.
Forty five
patients; mean
age 62.1 yrs
11.5; 56% male
Anthropometric
measures,
biochemical
markers, MUST,
NRI, & SGA
Compared to
prealbumin,
sensitivity,
specificity, PPV,
NPV, kappa:
MUST - 64%,
60%, 66.7%,
57.1%, 0.239;
NRI - 80.9%,
70.8%, 70.8%,
80.9%, 0.51;
SGA - 72.4%,
81.2%, 87.5%,
69.1%, 0.50
Overall the NRI had
the highest
sensitivity and
second highest
specificity when
compared to
prealbumin than the
MUST and SGA
tools. The MUST
and NRI tools were
comparable
measures, easy to
administer and
require minimal
training to complete,
compared to the
SGA. MUST is best
to use.
Small sample
size and
convenience of
inclusion. Not
many patient
demographics
noted in the
study.
31
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographic
s
Nutrition
Screening
Methods
Results Conclusions Limitations
Van
Venrooij,
Van
Leeuwen,
Hopmans,
Borgmeije
r-Hoelen,
De Vos, &
De Mol:
2011
Single-center
prospective
observational
cohort study
over 23
months.
Purpose was to
compare
undernutrition
screening tools
to low-fat free
mass index in
patients
undergoing
cardiac
surgery, and
assess
association
with
postoperative
adverse
outcomes.
Three
hundred
twenty five
adult cardiac
surgery
patients;
mean age
65.7 10.1;
27.7%
female;
19.4% had
BMI > 30;
4% had BMI
< 21
low-fat free
mass index
(FFMI),
MUST,
SNAQ,
cardiac
surgery-
specific
version of
MUST
(CSSM)
Accuracy in
detecting FFMI,
prevalence,
sensitivity,
specificity, PPV,
NPV, positive
likelihood ratio, area
under the curve:
MUST - 8.3%,
59.3%, 82.7%,
23.9%, 95.7%, 3.4,
0.71; SNAQ - 8.3%,
18.5%, 93.6%,
20.8%, 92.6%, 2.9,
0.56; CSSM - 8.3%,
74.1%, 70.1%,
18.5%, 96.7%, 2.5,
0.72. Post-op
adverse outcomes
defined by MUST &
SNAQ: 5.8%
infection, 2.5%
mortality, 36.4%
prolonged ICU LOS,
33.1% prolonged
hospital LOS.
Accuracy in
detecting FFMI
before surgery was
considerably higher
for MUST than
SNAQ. SNAQ
does not identify
'unintentional
weight loss' which
is important in
determining
malnutrition risk.
Further research on
the cardiac specific
MUST is
recommended
because it
integrates age and
sex. It is
recommended to
use the FFMI
measure with
unintentional
weight loss and low
BMI to identify and
refer malnourished
patients.
The bioelectrical
impedance makes
assumptions and
therefore the true
nutritional status
may be affected by
disease state. In
cardiac patients
bioelectrical
impedance can be
affected by higher
BMIs and
extracellular fluid
imbalances. The
reference standard
for undernutrition
does not take into
account weight
loss and low BMI.
BMI is only a
blunt tool for
measuring body
fatness. Experts
lack agreement
about an optimal
definition and
operationalism of
undernutrition.
32
Table 1 (continued)
Author and
Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Velasco,
Garcia,
Rodriguez,
Frias,
Garriga,
Alvarez,
Peris, &
Leon: 2010
Observational
multicenter
study to evaluate
nutritional risk in
hospitalized
patients using
four screening
tools.
Evaluations
performed by a
single
investigator at
each hospital
over five
months.
Four hundred
patients; mean
age 67.4 16.1
yrs; 60% male;
66% medical,
34% surgical
patients. Main
diagnoses were
pneumonia, HF,
COPD, surgery,
neurologic
vascular disease,
and other medical
diagnoses.
NRS
2002,
MUST,
SGA, &
MNA
Compared to
SGA,
sensitivity,
specificity, PPV,
NPV,
agreement: NRS
2002 - 74.4%,
87.2%, 76.1%,
86.2%, 0.62;
MUST - 71.6%,
90.3%, 80.1%,
85.4%, 0.64;
MNA - 95%,
61.3%, 57.2%,
95.7%, 0.491.
LOS for patients
(p<0.001): No
risk - NRS 2002
8.9days, MUST
9.2days, MNA
8.1days, SGA
8.8days; At risk
- NRS 2002
13.7days,
MUST 13.6days,
12.4days, SGA
13.7days.
Best agreement
with MUST and
SGA, and NRS-
2002 and SGA.
It is
recommended to
use MUST and
NRS-2002 upon
admission. MNA
detected more
patients at risk
but it can only be
used in the
elderly
population.
Some patients who
could not be
weighed gave an
estimation of their
weight, which
could lead to
skewed results.
There was a lower
prevalence of
malnourished
patients compared
to similar studies
in surgical
populations. This
may be due to the
fact that this study
mainly comprised
of elective
surgeries where
patients may be in
better nutritional
health.
33
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Vicente,
Barao,
Silva, &
Forones:
2013
Cross-
sectional
study to
evaluate
nutritional
screening
methods used
to screen
patients seen
in an
oncology
clinic in Sao
Paulo during
an 18 month
period.
One hundred thirty
seven colorectal
(n=116) and gastric
(n=21) cancer
patients divided into
two groups; group
one undergoing
treatment for cancer,
mean age 60.2
12.2yrs, 48% male;
group two patients
post tumor removal
undergoing follow-
up treatment, mean
age 61.3 11.6 yrs,
45.2% male
BMI,
albumin,
SGA,
NRI,
MUST
and MST
Compared to
SGA; Group 1
sensitivity,
specificity: BMI
- 10%, 100%;
albumin - 30%,
92%; NRI - 68%,
64%; MST -
52%, 84%;
MUST - 72%,
49%. Group 2
sensitivity,
specificity: BMI
- 15.3%, 100%;
albumin -
15.3%, 93.8%;
NRI - 55.8%,
83.6%; MST -
61.5%, 91.8%;
MUST - 84%,
73.4%.
MUST was the
most sensitive tool
for screening
nutrition, but with
a lower specificity.
NRI had a lower
sensitivity but a
higher specificity.
Overall the
subjective tools
showed a higher
sensitivity but
lower specificity
then objective
measures. MUST
and SGA in
combination are
better for
identifying
nutritional risk.
Although the
sample size was
large, it included a
small number of
patients with gastric
cancer, only 15% of
the study
population. The
authors noted
inconsistency with
other studies in the
number of
malnutrition patients
compared to other
studies in similar
populations. This
was attributed to the
patients not being
hospitalized and in
fairly good health.
34
References
Almeida, A. I., Correia, M., Camilo, M., & Ravasco, P. (2011). Nutritional risk
screening in surgery: Valid, feasible, easy! Clinical Nutrition, 31, 206-211.
Amaral, T. F., Antunes, A., Cabral, S., Alvest, P., & Kent-Smith, L. (2008). An
evaluation of three nutritional screening tools in a Portuguese oncology centre.
Journal of Human Nutrition and Dietetics, 21, 575-583.
Boleo-Tome, C., Monteiro-Grillo, I., Camilo, M., & Ravasco, P. (2011). Validation of
the Malnutrition Universal Screening Tool (MUST) in cancer. British Journal of
Nutrition, 108, 343-348.
Centers for Disease Control and Prevention (2013). Heart failure facts. Retrieved from:
http://www.cdc.gov/dhdsp/data_statistics/fact_sheets/docs/fs_heart_failure.pdf
Corkins, M. R., Guenter, P., DiMaria-Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S.,
… Resnick, H. E. (2014). Malnutrition diagnoses in hospitalized patients: United
States, 2010. Journal of Parenteral and Enteral Nutrition, 38, 186-195.
Elia, M. (2009). The economics of malnutrition. Nestle Nutrition, 12, 29-40.
Elia, M. & Malnutrition Advisory Group (2010). The ‘MUST’ Report: Nutritional
screening of adults, a multidisciplinary responsibility. British Association of
Parenteral and Enteral Nutrition, 1-11.
Gibson, S., Sequeira, J., Cant, R., & Ku, C. (2012). Identifying malnutrition risk in acute
medical patients: Validity and utility of Malnutrition Universal Screening Tool
and Modified Malnutrition Screening Tool. Nutrition and Dietetics, 69, 309-314.
Go, A. S., Mozaffarian, D., Roger, V. L., Benjamin, E. J., Berry, J. D., Blaha, M. J. …
Turner, M. B. (2013). Heart disease and stroke statistics – 2014 update. A report
35
from the American Heart Association. Circulation, 129, e28-e292.
Gout, B. S., Barker, L. A., & Crowe, T. C. (2009). Malnutrition identification, diagnosis
and dietetic referrals: Are we doing a good enough job? Nutrition and Dietetics,
66, 206-211.
Hoes, M. (2007). Clinical epidemiology of heart failure. Heart, 93, 1137-1146.
Jensen, G. L., Mirtallo, J., Compher, C., Dhaliwal, R., Forbese, A., Grijalba, R. F., …
Waitzberg, D. (2010). Adult starvation and disease-related malnutrition: A
proposal for etiology-based diagnosis in the clinical practice setting from the
International Consensus Guideline Committee. Journal of Parenteral and Enteral
Nutrition, 34(2), 156-159.
Kyle, U. G., Kossovsky, M. P., Karsegard, V. L. & Pichard, C. (2006). Comparison of
tools for nutritional assessment and screening at hospital admission: A population
study. Clinical Nutrition, 25, 409-417.
Lawson, C. S., Campbell, K. L., Dimakopoulos, I., & Dockrell, M. E. (2012). Assessing
the validity and reliability of the MUST and MST nutrition screening tools in
renal inpatients. Journal of Renal Nutrition, 22(5), 499-506.
Leistra, E., Langius, J. A., Evers, A. M., van Bokhorst-de van der Schueren, M. A.,
Visser, M., de Vet, H. C., & Kruizenga, H. M. (2013). Validity of nutritional
screening with MUST and SNAQ in hospital outpatients. European Journal of
Clinical Nutrition, 67, 738-742.
Lomivorotov, V. V., Efremov, S. M., Boboshko, V. A., Nikolaev, D. A., Vedernikov, P.
E., Lomivorotov, V. N., & Karaskov, A. M. (2012). Evaluation of nutritional
screening tools for patients scheduled for cardiac surgery. Nutrition, 29, 436-442.
36
Moughrabi, S. M. & Evangelista, L. S. (2007). Cardiac cachexia at a glance. Progress in
Cardiovascular Nursing, Spring, 101-103.
Neelemaat, F., Meijers, J., Kruizenga, H., van Ballegooijen, H., & van Bokhorst-de van
der Schueren, M. (2011). Comparison of five malnutrition screening tools in one
hospital inpatient sample. Journal of Clinical Nursing, 20, 2144-2152.
Nicol, S. M., Carroll, D. L., Homeyer, C. M., & Zamagni, C. M. (2002). The
identification of malnutrition in heart failure patients. European Journal of
Cardiovascular Nursing, 1, 139-147.
Olivares, J., Ayala, L., Salas-Salvado, J., Muniz, M. J., Gamundi, A., Martinez-Indart, L.,
& Masmiquel, L. (2014). Assessment of risk factors and test performance on
malnutrition prevalence at admission using four different screening tools.
Nutricion Hospitalaria, 29(3), 674-680.
Poulia, K. A., Yannakoulia, M., Karageorgou, D., Gamaletsou, M., Panagiotakos, D., B.,
Sipsas, N., V., & Zampelas, A. (2012). Evaluation of the efficacy of six
nutritional screening tools to predict malnutrition in the elderly. Clinical
Nutrition, 31, 378-385.
Stratton, R. J., Hackston, A., Longmore, D., Dixon, R., Price, S., Stroud, M., …Elia, M.
(2004). Malnutrition in hospital outpatients and inpatients: prevalence,
concurrent validity and ease of use of the ‘Malnutrition Universal Screening
Tool’ (MUST) for adults. British Journal of Nutrition, 92, 799-808.
Tu, M. Y., Chien, T. W., & Chou, M. T. (2012). Using a nutritional screening tool to
evaluate the nutritional status of patients with colorectal cancer. Nutrition and
Cancer, 64(2), 323-330.
37
Van Venrooij, L. M., Van Leeuwen, P. A., Hopmans, W., Borgmeijer-Hoelen, M. M., De
Vos, R., & De Mol, B. A. (2011). Accuracy of quick and easy undernutrition
screening tools – Short Nutritional Assessment Questionnaire, Malnutrition
Universal Screening Tool, and Modified Malnutrition Universal Screening Tool –
in patients undergoing cardiac surgery. Journal of the American Dietetics
Association, 111, 1924-1930.
Velasco, C., Garcia, E., Rodriguez, V., Frias, L., Garriga, R., Alvarez, J., … Leon, M.
(2010). Comparison of four nutritional screening tools to detect nutritional risk in
hospitalized patients: a multicenter study. European Journal of Clinical
Nutrition, 65, 269-274.
Vicente, M. A., Barao, K., Silva, T. D., & Forones, N. M. (2013). What are the most
effective methods for assessment of nutritional status in outpatients with gastric
and colorectal cancer? Nutricion Hospitalaria, 28, 585-591.
38
Manuscript 2
Nutritional Risk Index: A Review of the Literature
Cassondra Degener RD, BSN, RN, CCRN, DNP Student
University of Kentucky
39
Abstract
Purpose: The purpose of this literature review was to analyze the available literature on
the Nutritional Risk Index (NRI) tool and provide evidence to support its reliability and
validity in various populations. The literature review will examine the strengths and
weaknesses of NRI compared to other nutrition evaluation methods, compare results
between studies and make recommendations for practice.
Design and Methods: Literature review for articles in English or translated into English
from 2004 to 2014 using the following databases: EBSCOhost, CINAHL, MEDLINE and
Academic Search Complete. Once articles were chosen to be included in the review,
footnote chasing took place to find additional studies which evaluated NRI and other
screening methods.
Results: Over 34 articles were found but ten studies which met the inclusion criteria of
NRI and other screening methods. Of the included studies, three evaluated hospital
inpatients, three screened cancer patients, two assessed the nutritional status of the
elderly, and two examined the nutritional status of HF patients.
Practice Implications: Implementation of the NRI on admission in combination with
anthropometric measures may help assist providers in identifying multiple patient
populations at risk for malnutrition. No one tool has been proven as the gold standard of
nutrition assessment, making it necessary to evaluate multiple tools in the HF population.
Search Terms: malnutrition, screening tools, adult, nutritional risk index, hospitalized
patients, heart failure, cancer, surgery, chronic disease and elderly
40
Background and Significance
Malnutrition is associated with increased healthcare costs and worse outcomes
among hospitalized patients (Elia, 2009). According to the most current nationally-
representative data describing US hospital discharges, the average patient remains in the
hospital for 4.4 days, while malnourished patients spend an average of 12.6 days (Corkins
et al., 2014). The longer hospital stay triples healthcare costs for malnourished patients,
increasing from $9,485 for the average patients to $26,944 for malnourished ones
(Corkins et al., 2014). Malnutrition is present when a patient’s serum albumin level is
less than 3.3g/dL, the transferrin is less than 0.16g/dL, and/or the prealbumin is less than
15mg/dL (Beck & Rosenthal, 2002). Malnourished patients have increased hospital
length of stay (LOS) and increased readmission rates and are more likely to be discharged
to a long term care or rehabilitation facility (Chima et al., 1997).
Heart failure is one of the leading causes of death in the United States, and in
2009 one in nine deaths included HF as a contributing cause (Centers for Disease Control
and Prevention [CDC], 2013). The CDC estimates 5.1 million people in the United
States have HF (2013). Malnutrition is highly prevalent among hospitalized HF patients
at a rate of 66%, but often remains unidentified and untreated (Aziz et al., 2011; Stratton
et al., 2006). Most HF patients are unable to consume enough calories to meet the body’s
demands, which often leads to a condition called cardiac cachexia (Nicol et al., 2002).
Cardiac cachexia is a disorder characterized by muscle wasting and protein-energy
malnutrition (Moughrabi & Evangelista, 2007). A patient with HF who loses 7.5% or
more of his or her body weight over a period of six months most likely has cardiac
cachexia (Anker et al., 1997). Cardiac cachexia is responsible for increased morbidity
41
and mortality and decreased quality of life among patients with HF (Moughrabi &
Evangelista, 2007). There is a growing need to find a standardized tool to help providers
identify malnourished patients earlier and intervene faster. Early recognition of
malnutrition by healthcare providers could lead to early intervention, decreased morbidity
and mortality, and decreased healthcare costs and LOS (Elia, 2009; Stratton, Green, &
Elia, 2004).
Researchers have developed a number of different screening tools to assist
healthcare providers with the identification of malnutrition in hospitalized patients. One
tool in particular, the Nutritional Risk Index (NRI), was developed by the Veterans’
Affairs Total Parenteral Nutrition Cooperative Study Group to determine nutritional risk
in the postsurgical patient population (Al-Najjar & Clark, 2012). The NRI uses objective
measurements to calculate a score from the following formula: 1.5 x serum albumin +
41.7 x current weight/ideal body weight (Aziz et al., 2011). A score of > 100 means
there is no evidence of malnutrition, 97.5 – 100 indicates mild malnutrition, 83.5 – 97.5
means moderate malnutrition, and < 83.5 signifies severe malnutrition (Al-Najjar &
Clark, 2012). Since its development, the tool has demonstrated evidence of validity in
many patient populations including hospitalized patients, outpatients, surgical patients,
the elderly and those with HF and cancer, making it useful to implement in any setting or
population (Al-Najjar & Clark, 2012; Faramarzi, et al., 2013; Almeida et al., 2011).
Purpose of the Integrated Review
The purpose of this literature review was to analyze the available literature on the
NRI and provide evidence to support whether or not it is reliable in various populations
(i.e. oncology, hospitalized, elderly and HF). The following review evaluates ten studies
42
which compared NRI to other reliable screening tools such as Subjective Global
Assessment (SGA), Nutrition Risk Screening 2002 (NRS 2002), Mini-Nutritional
Assessment (MNA) and MUST in multiple patient populations. As of now, no one tool
has been shown to be a gold standard for evaluating nutritional status among all patient
populations in the different healthcare settings. The literature review will examine the
strengths and weaknesses of the NRI compared to other nutrition evaluation methods,
compare results between studies as to which screening tools are the most reliable and
make recommendations for practice.
Methods
Search Method
A search for published studies comparing NRI to other reliable nutritional
screening instruments and methods was executed using EBSCOhost via the UK Libraries
website. Databases used within EBSCOhost included Academic Search Complete,
CINAHL, and MEDLINE. The search only included articles in English or translated into
English ranging from 2004 to 2014. Keywords used in the search were malnutrition,
screening tools, adult, nutritional risk index, hospitalized patients, heart failure, cancer,
surgery, chronic disease and elderly. Once articles were chosen to be included in the
review, I used ancestry searching to find additional studies which evaluated NRI and
other screening methods. Inclusion criteria involved recent studies that compared NRI to
other screening methods. Studies were excluded if they were not published in English,
were written before 2004, or did not compare NRI to other reliable tools or screening
methods.
43
Search Outcome
The overall search produced 34 articles, of which ten met the criteria of
comparing NRI to other reliable screening tools and objective measures (albumin, weight
loss and BMI). Of the included studies, three evaluated hospital inpatients, three
screened cancer patients, two assessed the nutritional status of the elderly, and two
evaluated HF patients. The designs of the studies varied ranging from three prospective,
three cross-sectional, two controlled population, one retrospective cohort and one
retrospective analysis.
Findings and Synthesis of Themes
There were major themes that arose from this review which related to
malnutrition risk screening and the best methods in which to do so upon healthcare
admission. This review included ten studies which compared NRI to other reliable
screening methods in order to evaluate the nutritional status of various patient
populations. Screening tools evaluated in addition to NRI in the studies included: NRS-
2002, MUST, SGA, Patient Generated SGA (PG-SGA), Geriatric Nutritional Risk Index
(GNRI), Mini-nutritional Assessment Short Form (MNA-SF), and Malnutrition
Screening Tool (MST).
Of the ten studies, three recommended specific screening tools as the most
reliable methods of nutritional screening. Kyle, Kossovsky, Karsegard, and Pichard
(2006) recommended NRS-2002, MUST and SGA, while Aziz et al. (2011), and Al-
Najjar and Clark (2012) suggested the use of NRI. Six studies found the best methods
were a subjective screening tool in combination with anthropometric measures or
objective laboratory values. Of those six studies, MUST and SGA were suggested as the
44
subjective measures of choice by Poulia et al. (2012) and Vicente, Barao, Silva and
Forones (2013), while Tu, Chien and Chou (2012) suggested MUST and NRI. Three
studies did not mention specific subjective screening tools to be used in combination with
objective measures (Faramarzi, Mahdavi, Mohammad-Zadeh and Nasirimotlagh, 2013;
Meireles, Wazlawik, Bastos and Garcia, 2012; & Cereda, Limonta, Pusani, and Vanotti,
2006). The final study simply recommended recent percentage of weight loss as the
minimal screening method, even though the authors found NRS-2002 and MUST were
the most concordant, reliable and valid tools to use in surgical patients (Almeida, Correia,
Camilo, & Ravasco, 2011).
According to all the authors, they did agree that the best method for nutritional
screening is the use of an easy and inexpensive tool that requires little training. A few
authors suggested more research be done to determine the best all-around screening tool
to use in multiple patient populations, noting the inconsistency among current literature
and screening tools (Al-Najjar & Clark, 2012; Aziz et al., 2011; Cereda et al., 2006). The
following sections will illustrate those themes. Refer to Table 1 for specifics about each
study and their limitations.
Recommendation of Specific Subjective Tools
Two studies that reviewed NRI and HF were conducted by Al-Najjar and Clark
(2012) and Aziz et al. (2011). In their study, Al-Najjar and Clark (2012) included
outpatients with left ventricular chronic HF attending a community HF clinic, while Aziz
et al. (2011) evaluated serum albumin and NRI to assess the incidence of malnutrition
and outcomes of adults admitted to the hospital with acute decompensated HF. Al-Najjar
and Clark (2012) found NRI to be a univariable predictor of mortality (chi-square 25,
45
p<0.001), and an independent predictor of outcomes in multivariable analysis (chi-square
12, p<0.001). Aziz et al. (2011) determined NRI was the strongest predictor for LOS
(odds ratio 1.7, 95% confidence interval 1.58-1.9; p=0.005). The authors also found
moderate to severe NRI scores were associated with higher death and readmission rates
(Aziz et al., 2011). Based on their statistical analysis, the authors of both studies
concluded the NRI was a helpful prognostic marker in patients with HF compared to BMI
or albumin alone (Al-Najjar & Clark, 2012; Aziz et al., 2011). The authors also
recognized the need for more randomized controlled studies which evaluate NRI and HF
patients in order to find a consistent and reliable screening method in this population (Al-
Najjar & Clark, 2012; Aziz et al., 2011).
The other study which recommended specific subjective screening tools evaluated
the nutritional status of hospitalized adult medical and surgical patients (Kyle et al.,
2006). The authors used SGA as the gold standard and compared results between three
other tools, NRI, MUST, and NRS-2002 (Kyle et al., 2006). They found MUST and
NRS-2002 to be the most concordant with SGA with kappa values of 0.26 and 0.48
respectively. The MUST had the advantage of being less time consuming and required
less examiner training, even though it produced a lower sensitivity and specificity of
61.2% and 78.6% respectively (Kyle et al., 2006). Based on statistical analysis and the
tool’s ease of use, the authors concluded that NRS-2002, SGA, and to a lesser extent
MUST, were the best screening tools to evaluate patients upon hospital admission (Kyle
et al., 2006).
46
Combination of Objective and Subjective Screening Methods
Poulia et al. (2011) evaluated the efficacy of six subjective screening tools to
predict malnutrition in hospitalized elderly patients (>60 years old) admitted to the
hospital in Athens, Greece. Tu et al. (2011) used three screening tools to evaluate the
nutritional status of colorectal cancer patients, while Vicente et al. (2013) used four tools
to screen colorectal and gastric cancer patients. Poulia et al. (2011) found MUST and
MNA-SF to be the most reliable with a sensitivity of 87.3% and 98.1%, and specificity of
76.8% and 50% respectively. The best agreement with the combined index (gold
standard) was with SGA and MUST noting kappa values of 0.71 and 0.64 respectively
(Poulia et al., 2011). In their study, Tu et al. (2011) found MUST, NRI and SGA to have
sensitivities of 64%, 80.9%, and 72.4% with specificities of 66.7%, 70.8%, and 81.2%
respectively. In contrast to SGA, the authors determined MUST and NRI were
comparable measures, easy for healthcare providers to administer, and required minimal
training to complete (Tu et al., 2011). Vicente et al. (2013) determined MUST had a
sensitivity of 84% and specificity of 73.4% when compared to SGA. They also found the
subjective measures to have higher sensitivities but lower specificities than the objective
measures (Table 1). Based on statistical analysis, the authors suggested that MUST and
SGA were the best screening measures (Poulia et al, 2011; & Vicente et al., 2013). Tu et
al. (2011) on the other hand recommended the use of MUST and NRI when screening
hospitalized cancer patients due to their ease of use and requirement of minimal training
to complete.
Cereda et al. (2006) compared NRI and GNRI to albumin and prealbumin in
elderly patients admitted to a long-term care facility in Como, Italy. Faramarzi et al.
47
(2013) screened colorectal cancer patients using albumin, NRI and PG-SGA, while
Meireles et al. (2012) screened hospitalized surgical patients using three screening tools
and anthropometric measures (fat mass index, body cell mass, and standardized phase
angle). Cereda et al. (2006) found NRI and GNRI to have similar reliability using
Pearson’s linear correlation coefficients of 0.98 and 0.95 respectively, in comparison to
the objective measures of albumin and prealbumin. Faramarzi et al. (2013) found NRI to
have a sensitivity, specificity and kappa value of 66%, 60% and 0.27 when compared to
PG-SGA. Using SGA as the gold standard, Meireles et al. (2012) found NRS-2002 and
NRI had kappa coefficient values of 0.49 and 0.26 respectively. Based on their
statistical analysis, the authors suggested a combination of subjective and objective
measures, but did not recommend a specific screening tool (Cereda et al., 2006;
Faramarzi et al., 2013; & Meireles et al., 2012). Cereda et al. (2006) further suggested
the need for long-term prospective studies which evaluate the nutritional status of the
elderly using multiple subjective screening tools and objective measures.
Objective Measures Recommended for Practice
Almeida et al. (2011) conducted their study on 300 adult surgical patients
admitted to the hospital. They used the Subjective Global Assessment (SGA) as the gold
standard nutritional screening method and compared it to three subjective screening tools,
BMI and percentage weight loss. They found MUST and NRS-2002 to be the most
concordant with SGA (Almeida et al., 2011). The sensitivity of MUST and NRS-2002
were 85% and 80%, while the specificities were 89% and 93% respectively. The
sensitivity and specificity of percentage weight loss were also higher at 89% and 93%
respectively. Compared to the subjective measures, percentage weight loss was more
48
cost effective and less time consuming (Almeida et al., 2011). The authors concluded
that percentage weight loss screening on admission should be mandatory in routine
practice at the very least, in order to increase outcome driven nutrition management
(Almeida et al., 2011).
Practice Implications
Early identification and treatment of malnutrition may help decrease hospital
costs, inpatient LOS and readmission rates. The costs of HF alone are high. The
presence of a complication such as malnutrition can increase healthcare costs and the
length of hospital stays dramatically. The NRI has demonstrated evidence of reliability
and validity in the hospitalized, cancer, elderly, and HF patient populations. Best results
are seen when a subjective tool is used in combination with anthropometric (BMI and %
weight loss) and laboratory measures (serum albumin and prealbumin) to identify those at
risk for malnutrition. Implementation of subjective screening tools on admission in
combination with anthropometric measures may help assist providers in identifying
multiple patient populations at risk for malnutrition.
There are major gaps in the literature in regards to consistency among nutrition
screening tools in multiple populations. There are conflicting data in research today as to
which nutritional screening tool is the most valid and reliable across various clinical
settings and in different patient populations. The elderly, cancer and hospitalized patients
were the most common populations in which NRI was evaluated; however, in those
studies, researchers mainly evaluated NRI and SGA.
Two of the ten studies recommended the use of NRI alone while the others which
found NRI useful, also recommended other screening tools. In the two studies which
49
evaluated NRI and HF patients, the tool was not compared to other subjective screening
tools, only objective and anthropometrics measures. It would be helpful to see NRI
compared to other similar screening tools to determine the most reliable in HF patients.
Overall, there needs to be more studies in which a variety of tools are used in accordance
with biometric nutritional screening parameters such as laboratory data and body
measurements in order to determine the most accurate and reliable screening tool.
Conclusion
This literature review revealed a lack of studies in which multiple tools evaluated
the nutritional status of HF patients. Of the two studies reviewed that pertained to HF
patients, one study compared NRI to traditional nutritional biomarkers, while the second
used NRI to evaluate HF patient outcomes. Multiple subjective screening tools need to
be studied within this population to better identify malnutrition among HF patients. No
one tool has been proven as the gold standard of nutrition assessment, making it
necessary to evaluate multiple tools in the HF population.
50
Table 1 Review of the NRI Literature
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Al-Najjar,
& Clark:
2012
Retrospective
cohort study
over six years at
a community
CHF clinic.
Evaluate
nutrition
screening
methods and
their application
to HF patients.
Five hundred
thirty eight
outpatients;
mean age 71
9.9; 76% male;
all with left
ventricular
systolic
dysfunction
Nutritional
Risk Index
(NRI), BMI
and various
laboratory
variables
(albumin,
hemoglobin,
white blood
count,
platelets,
creatinine,
potassium)
Prevalence
according to
NRI: 23%
moderate, 2.8%
being severe.
NRI correlation
coefficient:
BMI 0.87,
hemoglobin 0.19, and age -
0.24 (p < 0.001)
Increased age and
decreased BMI and
hemoglobin were
associated with
increased incidence
of malnutrition.
NRI was a helpful
prognostic marker
in patients with HF
in an outpatient
setting. There is a
need for a large
randomized
controlled trials
using NRI to
evaluate
malnutrition effects
on mortality.
Large study
population but did
not take into
account changes in
medical therapy
for patients when
determining
malnutrition
prevalence. The
patient population
was a convenience
sample and
included a large
number of males
compared to
females, 76% and
24% respectively.
They also did not
compare NRI to
other methods of
nutritional analysis
in determining
prevalence of
malnutrition in the
HF population.
51
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Almeida,
Correia,
Camilo, &
Ravasco:
2011
Prospective
cross-sectional
study, over
eight months
with all data
collection by a
single research
dietitian to
determine
nutritional
status
Three hundred
surgical
hospitalized
patients; ages
43 - 77; 44%
male; 46%
cancer patients
BMI, recent
% weight
loss,
Nutrition
Risk
Screening
2002 (NRS
2002),
Malnutrition
Universal
Screening
Tool
(MUST),
NRI,
Subjective
Global
Assessment
(SGA)
Compared to
SGA the
Sensitivity,
Specificity,
positive
predictive value
(PPV), negative
predictive value
(NPV): NRS
2002 - 80%,
89%, 87%,
100%; MUST -
85%, 93%,
89%, 99%; NRI
- 29%, 27%,
24%, 27%;
BMI - 43%,
39%, 35%,
31%; % wt loss
- 89%, 93%,
81%, 89%
NRS 2002 and
MUST are the most
concordant, valid
and reliable tools to
detect nutrition risk
in surgical patients.
>5% weight loss
over six months was
reliable and valid.
Percent weight loss
estimation should
be mandatory in
routine practice to
increase outcome
driven nutrition
management.
MUST and NRI
were made into
two categories for
the purpose of the
study, but each
were originally
developed into
three and four
categories. Made
two categories in
order to determine
comparisons, but
only two
categories could
affect the results.
52
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Aziz,
Javed,
Pratep,
Musat,
Nader,
Pulimi,
Alivar,
Herzog &
Kukin:
2011
Controlled
population
study to assess
the incidence of
malnutrition
and outcomes
of adults
admitted with
acute
decompensated
HF
One thousand
one hundred
patients with
acute
decompensated
HF; No risk
mean age 68
14yrs, 51%
male; mild risk
mean age 72
14yrs, 51%
male; moderate
risk 72 14yrs,
59% male;
severe risk
mean age 68
15yrs, 56%
male
Serum
albumin and
NRI
NRI risk: none
666 (60%),
mild 63 (6%),
moderate 213
(19%), severe
168 (15%).
Values for
mod/severe
risk:
readmission
rates
52%/68%; LOS
10/10.9 days;
mortality rates 15%/19% (p <
0.001). Average
albumin: no risk
3.9, mild risk
3.4, mod risk
3.2, severe 2.6
NRI scores
correlated with the
lower serum
albumin levels.
NRI is better
prognostic indicator
of morbidity and
mortality in HF
patients than BMI
or albumin alone.
Recommend NRI to
further stratify these
patients for nutrition
depletion
assessment. Need
more trials to
determine if
nutrition therapy is
helpful to improve
outcomes in HF
patients.
Could not
calculate periodic
NRI scores in
patients after
admission to the
hospital. This
could have helped
evaluate
nutritional status
throughout the
hospital stay
which may have
affected patient
outcomes.
53
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Cereda,
Limonta,
Pusani, &
Vanotti:
2006
Retrospective
analysis to
compare
traditional
malnutrition
markers and
screening tools
to determine
malnutrition
risk of elderly
admitted to a
long-term care
unit
One hundred
seventy seven
elderly patients;
38% male;
mean age 80
8.6 yrs
Albumin,
prealbumin,
Geriatric
Nutritional
Risk Index
(GNRI) and
NRI
Nutrition risk:
GNRI - mod
14.2%, severe
3.5%; NRI -
mod 33.8%,
severe 3.9%.
Pearson's linear
correlation
coefficient:
albumin -
GNRI 0.95;
NRI 0.98;
prealbumin -
GNRI 0.52,
NRI 0.52
Concluded that a
prospective study
comparing the two
tools would be
beneficial given
their similar
reliability and
agreement to
traditional markers
in elderly patients
admitted with an
acute illness. GNRI
and NRI showed
significant
correlations with all
other biochemical
markers of nutrition
status.
Sample included
patients picked for
convenience and
resided in a long-
term care setting.
Those patients are
usually less likely
to be at nutritional
risk as opposed to
those in the
hospital who are
acutely ill. The
study was
retrospective and
collected data
could be incorrect.
54
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Faramarzi,
Mahdavi,
Mohamma
d-Zadeh,
&
Nasirimotl
agh: 2013
Prospective
study to
validate NRI
against patient-
generated
subjective
global
assessment
(PG-SGA), in
adult colorectal
cancer patients
before
radiotherapy.
All data
collected by one
nutritionist.
Fifty two
patients; mean
age 54 years
16.8 yrs; 77%
male.
Anthropomet
ric data,
albumin,
NRI, PG-
SGA
Prevalence of
malnutrition:
SGA 33%
moderate, 19%
severe; NRI
35% mod, 10%
severe. When
compared to the
SGA, NRI
sensitivity 66%,
specificity 60%,
PPV 64%, NPV
62%, kappa
0.267
NRI had lower
sensitivity and
specificity than
SGA in assessing
nutritional status of
cancer patients.
Each tool has its
own advantages and
disadvantages (cost
and ease of use).
Best nutrition
assessment is a
combination of
anthropometric
measures and
subjective scoring
systems.
Small sample size
and convenience
of inclusion. NRI
tool uses albumin,
while SGA is
based on weight
history, dietary
intake, diagnosis
and physical
assessment.
Albumin may be
affected by disease
state and
inflammation,
making NRI
results less
accurate.
55
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Kyle,
Kossovsky
,
Karsegard
& Pichard:
2006
Population
study to test the
sensitivity and
specificity of
three screening
tools compared
to SGA, to
assess the
association
between
nutritional risk
and hospital
LOS over a
three month
period.
Nine hundred
ninety five adult
medical and
surgical
patients; 53%
male; mean age
50.5 21.9 (<
10d LOS), 65.4
18.7 (> 11d
LOS)
NRI, MUST,
NRS-2002,
SGA
Moderate/sever
e nutritional
risk: SGA 39%,
NRI 25%,
NRS-2002 28%, MUST
37%.
Compared to
SGA,
sensitivity,
specificity,
PPV, NPV,
kappa: NRI
43.1%, 89.3%,
76.2%, 66.3%,
0.24; MUST
61.2%, 78.6%,
64.6%, 76.1%,
0.26; NRS-
2002 62%,
93.1%, 85.1%,
79.4%, 0.48
Significant
association between
LOS and moderate
to severe
malnutritional status
among all tools.
NRS-2002 had
higher sensitivity
and specificity
compared to SGA
than NRI and
MUST. The
authors
recommended using
the NRS-2002,
MUST and SGA on
admission to screen
patients for
malnutrition.
SGA does not
allow for
categorization of
mild malnutrition
and focuses on
chronic not acute
malnutrition.
Screeners should
have been better
trained on the
screening tools
before
implementing the
study. LOS was
studied as an
outcome
parameter, but
many other factors
influence LOS, not
just malnutrition,
which were not
assessed in this
study.
56
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Meireles,
Wazlawik,
Bastos, &
Garcia:
2012
Cross-sectional
study to assess
the relationship
between
nutritional risk
tools and
parameters
derived from
bioelectrical
impedance
analysis with
SGA over 6
months.
One hundred
twenty four
hospitalized
surgical
patients; mean
age 52.26
14.95 yrs;
56.5% female;
33.1% > 60 yrs
SGA, NRS
2002, NRI,
Fat-Free
Mass Index
(FFMI), Fat
Mass Index
(FMI), body
cell mass
(%BCM),
standardized
phase angle
(SPA)
Nutritional risk:
NRS 2002 19.3%; NRI
69.5%; FFMI
12.9%; FMI
8.1%; %BCM
46.8%; SPA
4.8%.
Agreement
between SGA
and screening
parameters (k
coefficient):
NRS 0.490;
NRI 0.256;
FFMI 0.342;
FMI 0.190;
%BCM -0.085;
SPA 0.038
NRS 2002 showed
the best agreement
with SGA. Highest
malnutrition
prevalence seen
with NRI. The best
malnutrition
indicator is a
combination of
anthropometric
measures with
subjective screening
tools.
Sample size could
have been larger.
The BMI cut off
was 34. Obese
patients can be
very malnourished
and should have
been included in
the study.
57
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Poulia,
Yannakoul
ia,
Karageorg
ou,
Gamaletso
u,
Panagiotak
os, Sipsas,
&
Zampelas:
2012
Prospective
study to
evaluate the
efficacy of tools
to predict
malnutrition in
elderly patients
admitted to the
hospital in
Athens, Greece
over nine
months.
Two hundred
forty eight
elderly patients
> 60 years;
mean age 75.2
8.5yrs; 52%
male; admitted
for neurologic
syndrome,
fever, blood
disease,
rheumatologic
disease,
malignancy,
hemorrhage,
diabetes, GI,
kidney or
respiratory
disease
NRI, GNRI,
SGA,
MUST, mini
nutritional
assessment –
screening
form (MNA-
SF), and
NRS-2002
Compared to
true nutritional
status
(combined
index),
sensitivity,
specificity,
PPV, NPV,
kappa: NRI -
71.7%, 48.8%,
85.4%, 29.3%,
0.550; GNRI -
66%, 92.1%,
94.6%, 56.45%,
0.465; SGA -
84.3%, 91.4%,
95.2%, 74.3%,
0.707; MUST -
87.3%, 76.8%,
88.4%, 75%,
0.638; MNA-
SF - 98.1%,
50%, 79.9%,
93.2%, 0.545;
NRS 2002 -
99.4%, 6.1%,
68.2%, 83.3%,
0.088
The NRI was higher
in sensitivity and
PPV than the other
tools, but scored
lower in specificity
and NPV. MUST
and MNA-SF were
the most valid.
MUST and SGA
showed better
agreement with the
combined index.
The researchers
concluded that the
combination of
objective and
subjective
diagnostic tools that
are easy to use are
the best for
nutritional
screening.
Some patients had
to have the
questionnaires
translated for them
and results then
had to be
translated again,
making for some
miscommunication
among patients
and researchers.
Some patients
estimated height
and weight instead
of being accurately
measured by
researchers and in
4.8% of patients,
anthropometric
measurements
were not available.
These variations in
accuracy of
measurements
could affect
calculations and
results.
58
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Tu, Chien,
& Chou:
2012
Prospective
study to assess
the nutritional
status of
patients with
colorectal
cancer before
and after
surgery in
Taiwan over
two years.
Forty five
colorectal
cancer patients;
mean age 62.1
yrs 11.5; 56%
male
Anthropomet
ric measures,
biochemical
markers,
MUST, NRI,
& SGA
Compared to
prealbumin,
sensitivity,
specificity,
PPV, NPV,
kappa: MUST -
64%, 60%,
66.7%, 57.1%,
0.239; NRI -
80.9%, 70.8%,
70.8%, 80.9%,
0.51; SGA -
72.4%, 81.2%,
87.5%, 69.1%,
0.50
Overall the NRI had
the highest
sensitivity and
second highest
specificity when
compared to
prealbumin than
MUST and SGA.
The MUST and
NRI were
comparable
measures, easy to
administer and
require minimal
training to
complete, compared
to the SGA. MUST
is best to use.
Small sample size
and convenience
of inclusion
59
Table 1 (continued)
Author
and Year
Study Design
and Purpose
Sample and
Demographics
Nutrition
Screening
Methods
Results Conclusions Limitations
Vicente,
Barao,
Silva, &
Forones:
2013
Cross-sectional
study to
evaluate
nutritional
screening
methods used to
screen patients
seen in an
oncology clinic
in Sao Paulo
during an 18
month period.
137 colorectal
(n=116) and
gastric (n=21)
cancer patients
divided into two
groups; group
one undergoing
treatment for
cancer, mean
age 60.2
12.2yrs, 48%
male; group two
patients post
tumor removal
undergoing
follow-up
treatment, mean
age 61.3 11.6
yrs, 45.2% male
BMI,
albumin,
SGA, NRI,
MUST and
the
Malnutrition
Screening
Tool (MST)
Compared to
SGA; Grp 1
sensitivity,
specificity:
BMI - 10%,
100%; albumin
- 30%, 92%;
NRI - 68%,
64%; MST -
52%, 84%;
MUST - 72%,
49%. Grp 2
sensitivity,
specificity:
BMI - 15.3%,
100%; albumin
- 15.3%, 93.8%;
NRI - 55.8%,
83.6%; MST -
61.5%, 91.8%;
MUST - 84%,
73.4%.
MUST was the
most sensitive tool
for screening
nutrition, but with a
lower specificity.
NRI had a lower
sensitivity but a
higher specificity.
Overall the
subjective tools
showed a higher
sensitivity but lower
specificity then
objective measures.
MUST and SGA in
combination are
better for
identifying
nutritional risk.
Although the
sample size was
large, it included a
small number of
patients with
gastric cancer,
only 15% of the
study population.
The authors noted
inconsistency with
other studies in the
number of
malnutrition
patients compared
to other studies in
similar
populations. This
was attributed to
the patients not
being hospitalized
and in fairly good
health.
60
References
Al-Najjar, Y. & Clark, A. L. (2012). Predicting outcome in patients with left ventricular
systolic chronic heart failure using a Nutritional Risk Index. American Journal of
Cardiology, 109, 1315-1320.
Almeida, A. I., Correia, M., Camilo, M., & Ravasco, P. (2011). Nutritional risk
screening in surgery: Valid, feasible, easy! Clinical Nutrition, 31, 206-211.
Anker, S. D., Ponikowski, P., Varney, S., Chua, T. P., Clark, A. L., Webb-Peploe, K. M.,
…Coats, A. J. (1997). Wasting as independent risk factor for mortality in chronic
heart failure. The Lancet, 349(9058), 1050-1053.
Aziz, E. F., Javed, F., Pratep, B., Musat, D., Nader, A., Pulimi, S., … Kukin, M. L.
(2011). Malnutrition as assessed by Nutritional Risk Index is associated with
worse outcome in patients admitted with acute decompensated heart failure: An
ACAP-HF data analysis. Heart International, 6(2), 3-8.
Beck, F. K., & Rosenthal, T. C. (2002). Prealbumin: A marker for nutritional evaluation.
American Family Physician, 65(8), 1575-1578.
Center for Disease Control and Prevention (2013). Heart failure facts. Retrieved from:
http://www.cdc.gov/dhdsp/data_statistics/fact_sheets/docs/fs_heart_failure.pdf
Cereda, E., Limonta, D., Pusani, C., & Vanotti, A. (2006). Assessing elderly at risk of
malnutrition: The new Geriatric Nutritional Risk Index versus Nutritional Risk
Index. Nutrition, 22, 680-682.
Chima, C. S., Barco, K., Dewitt, M. A., Teran, J. C., & Mullen, K. D. (1997).
Relationship of nutritional status to length of stay, hospital costs, and discharge
status of patients hospitalized in the medicine service. Journal of the American
Dietetic Association, 97(9), 975-980.
61
Corkins, M. R., Guenter, P., DiMaria-Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S.,
… Resnick, H. E. (2014). Malnutrition diagnoses in hospitalized patients: United
States, 2010. Journal of Parenteral and Enteral Nutrition, 38, 186-195.
Elia, M. (2009). The economics of malnutrition. Nestle Nutrition, 12, 29-40.
Faramarzi, E., Mahdavi, R., Mohammad-Zadeh, M., & Nasirimotlagh, B. (2013).
Validation of Nutritional Risk Index method against Patient-Generated Subjective
Global Assessment in screening malnutrition in colorectal cancer patients.
Chinese Journal of Cancer Research, 25(5), 544-548.
Kyle, U. G., Kossovsky, M. P., Karsegard, V. L. & Pichard, C. (2006). Comparison of
tools for nutritional assessment and screening at hospital admission: A population
study. Clinical Nutrition, 25, 409-417.
Meireles, M. S., Wazlawik, E., Bastos, J. L., & Garcia, M. F. (2012). Comparison
between nutrition risk tools and parameters derived from bioelectrical impedance
analysis with Subjective Global Assessment. Journal of the Academy of Nutrition
and Dietetics, 112, 1543-1549.
Moughrabi, S. M. & Evangelista, L. S. Cardiac cachexia at a glance. Progress in
Cardiovascular Nursing, 2007; Spring: 101-103.
Nicol, S. M., Carroll, D. L., Homeyer, C. M., & Zamagni, C. M., (2002). The
identification of malnutrition in heart failure patients. European Journal of
Cardiovascular Nursing, 1, 139-147.
Poulia, K. A., Yannakoulia, M., Karageorgou, D., Gamaletsou, M., Panagiotakos, D., B.,
Sipsas, N., V., & Zampelas, A. (2012). Evaluation of the efficacy of six
nutritional screening tools to predict malnutrition in the elderly. Clinical
62
Nutrition, 31, 378-385.
Stratton, R. J., King, C., & Elia, M. (2004). Simple cost-analysis of changes in hospital
length of stay with oral nutritional supplements from randomized controlled trials.
Clinical Nutrition, 23, 906-915.
Stratton, R. J., King, C. L., Stroud, M. A., Jackson, A. A., Elia, M. (2006). 'Malnutrition
Universal Screening Tool' predicts mortality and length of hospital stay in acutely
ill elderly. British Journal of Nutrition 95(2), 325–330
Tu, M. Y., Chien, T. W., & Chou, M. T. (2012). Using a nutritional screening tool to
evaluate the nutritional status of patients with colorectal cancer. Nutrition and
Cancer, 64(2), 323-330.
Vicente, M. A., Barao, K., Silva, T. D., & Forones, N. M. (2013). What are the most
effective methods for assessment of nutritional status in outpatients with gastric
and colorectal cancer? Nutricion Hospitalaria, 28, 585-591.
63
Manuscript 3
Evaluating Nutritional Risk in Heart Failure Patients Using Four Screening Tools: A
Retrospective Chart Audit
Cassondra Degener RD, BSN, RN, CCRN, DNP Student
University of Kentucky
64
Abstract
The purpose of this project was to test the performance of albumin, NRI and MUST
screening tools in comparison to the standardized measure of prealbumin among HF
patients admitted to the University of Kentucky Chandler Medical Center. Inclusion
criteria included all HF patients with the 428 diagnostic code, admitted from January 1, -
December 31, 2013, ages 18 and older, with all laboratory values available specified in
the data collection tool (Appendix A). A retrospective electronic medical record (EMR)
review was performed for 100 patients who met the inclusion criteria. All data were
collected through the University of Kentucky’s secure network via the EMR program
Sunrise Clinical Manager. Serum albumin, prealbumin, MUST and NRI found 79, 85, 53
and 92 patients to be at nutritional risk, respectively. The NRI tool compared better with
prealbumin than albumin and MUST, when screening for malnutrition in HF patients.
The sensitivity of NRI compared to prealbumin was 92.9%. The results of this study
provide evidence that NRI in combination with laboratory measures may be beneficial in
identifying malnutrition among HF patients. There is still a need for further research into
effective screening methods among this population.
65
Evaluating Nutritional Risk in Heart Failure Patients Using Four Screening Tools: A
Retrospective Chart Audit
Malnutrition incidence among hospitalized inpatients is prevalent at a rate of 23%
(Gout, Barker, & Crowe, 2009). According to the most current nationally-representative
data describing U.S. hospital discharges, malnourished patients spent an average of 12.6
days in the hospital compared to 4.4 days for other patients (Corkins et al., 2014). With
an increased hospital length of stay, the average hospital cost will triple for those patients,
rising from $9,485 to $26,944 (Corkins et al., 2014). Visceral proteins such as albumin
and prealbumin are useful markers to detect malnutrition in adults and the elderly (Sergi
et al., 2006). Traditionally albumin has been the most commonly used indicator, with
prealbumin increasing in popularity in the recent years (Beck & Rosenthal, 2002).
Prealbumin is the most sensitive indicator for protein synthesis because it contains one of
the highest ratios of essential and nonessential amino acids compared to any protein in
the body (Beck & Rosenthal, 2002). Normal levels for albumin and prealbumin are 3.3-
4.8g/dL and 16-35mg/dL respectively (Beck & Rosenthal, 2002). Serum albumin has a
half-life of 20 days and can be affected by hydration status and renal function. The half-
life of prealbumin is two days and is not typically affected by hydration status, renal or
liver function. Prealbumin levels will decrease if a patient is consuming 60% or less of
their required daily protein intake (Le Moullac, Gouache, & Bleiberg-Daniel, 1992).
Once adequate supplementation of proteins is restored, increased prealbumin synthesis
will typically occur within 2-4 days (Le Moullac et al., 1992). Low levels of albumin and
prealbumin are associated with a low body mass index (BMI) and a poor nutritional
status (Sergi et al., 2006). In HF patients, renal insufficiency is common and can lead to
66
false elevation of serum albumin and prealbumin; therefore, malnutrition cannot be ruled
out if these levels are on the lower end of normal (Sergi et al., 2006).
An estimated 5.1 million Americans have HF (Centers for Disease Control, 2013).
Malnutrition prevalence among HF patients is as high as 66% based on serum albumin
levels and the presence of less than 90% ideal body weight (Aziz et al., 2011; Nicol et al.,
2002). Malnutrition in HF patients may be caused by hepatic and gastrointestinal (GI)
congestion due to elevated right sided heart pressures, resulting in anorexia,
malabsorption, dyspepsia, and protein wasting enteropathy (Nicol et al., 2002). These
changes may lead to the patient feeling full and satisfied due to hepatic and GI congestion
rather than consuming a full meal (Nicol et al., 2002). Cardiac cachexia (CC) is the
presence of severe malnutrition in HF patients which can be associated with advanced
myocardial dysfunction, poor prognosis and decreased survival (Moughrabi &
Evangelista, 2007). The definition of CC is “unintentional nonedematous weight loss
greater than 6% of a patient’s previous weight over a period of six months regardless of
BMI, and in the absence of other primary cachectic states such as cancer, thyroid disease
and severe liver disease” (Moughrabi & Evangelista, 2007, p. 101).
Changes in weight are not always an accurate measure of nutritional status given
the fluid volume overload often present in HF (Araujo, Lourenco, Rocha-Gonocalves,
Ferreira, & Bettencourt, 2011). Nutritional markers among patients with CC were
assessed, and prealbumin, albumin, hemoglobin, lymphocyte count and triglycerides
levels were significantly lower compared to healthy HF patients (Araujo et al., 2011).
Based on multivariate logistic regression analysis, prealbumin was the only laboratory
67
marker independently associated with CC occurrence through an odds ratio of 1.08 and
95% confidence interval 1.01-1.17 (p<0.001; Araujo et al., 2011).
Laboratory values of albumin and prealbumin are useful in identifying
malnutrition in the general and HF populations, but they have a few limitations in their
accuracy. Albumin concentrations can be affected by hydration, renal function, and the
presence of infection or inflammation (Beck & Rosenthal, 2002). Prealbumin can
decrease in the post-surgical phase, in the presence of inflammation, and in conditions
associated with protein malnutrition (e.g., malignancy, cirrhosis and zinc deficiency;
Beck & Rosenthal, 2002). Overall, prealbumin is a better nutrition laboratory marker of
acute changes in nutritional status, while albumin more accurately identifies chronic
malnutrition. Aside from prealbumin and albumin, there is currently a lack of literature
that compares other laboratory trends and trends in co-morbidities among malnourished
HF patients.
Subjective nutritional screening tools can be easy, rapid and inexpensive methods
of identifying malnutrition risk and prevalence among patients. There are a number of
screening tools available, but there are few studies which identify the best methods to
measure malnutrition and its severity in the HF population. However, two subjective
screening tools have shown some promise among this population, the Malnutrition
Universal Screening Tool (MUST) and the Nutritional Risk Index (NRI). In two studies,
multiple screening tools were tested in cardiac surgery patients and the MUST was
reported as being the most sensitive in detecting malnutrition (Lomivorotov et al, 2012;
Venrooij et al., 2011). In comparison to the Subjective Global Assessment (SGA)
screening tool, MUST had a sensitivity of 97.9 and specificity of 87.1 (Lomivorotov et
68
al., 2012). The MUST was also the only tool to be significantly associated with post-
operative complications following open heart surgery according to multivariate logistic
regression analysis (odds ratio 1.5; 95% confidence interval 1.1 – 2.4; p=0.02).
Researchers recommended MUST in screening cardiac surgery patients, but also
indicated more research needs to be conducted among HF patients to determine the most
reliable tool (Lomivorotov et al, 2012; Venrooij et al., 2011).
Two studies compared NRI to a traditional nutrition biomarker, albumin, to
determine its reliability in identifying malnutrition among HF patients (Al-Najjar &
Clark, 2012; Aziz et al., 2011). When evaluating NRI and other nutritional screening
parameters as predictors of outcomes and mortality, NRI was a useful prognostic marker
in outpatients with HF (Al-Najjar & Clark, 2012). According to statistical analysis, NRI
was a univariable predictor of mortality (chi-square 25, p <0.001), and an independent
predictor of outcome in multivariable analysis (chi-square 12, p <0.001; A-Najjar &
Clark, 2012). In another study, NRI was found to be the most significant predictor of all-
cause mortality and readmission rates associated with episodes of acute decompensated
HF (Aziz et al., 2011). Using Cox’s hazard regression models, NRI had a univariate odds
ratio of 3.03, and a 95% confidence interval of 3.22-3.94 with p < 0.0001; a multivariate
odds ratio of 3.1, and a 95% confidence interval of 2.34-4.22 with p <0.0001 (Aziz et al.,
2011). The authors suggested further research be conducted using NRI in the HF
population to determine malnutrition prevalence and its effects on morbidity and
mortality (Al-Najjar & Clark, 2012; Aziz et al., 2011). Implementing the use of NRI or
MUST on admission for HF patients may help identify the presence of malnutrition
earlier so that the malnourished may be referred to a dietitian for appropriate nutritional
69
intervention earlier. Noting the negative impacts of malnutrition on the patient, early
recognition and management may help decrease hospital lengths of stay, readmission
rates and associated healthcare costs.
Description of Practice Inquiry Project
This practice inquiry project, through a retrospective EMR review, evaluated the
presence of malnutrition in 100 HF patients admitted to the University of Kentucky
Chandler Medical Center between January 1, and December 31, 2013.
Goals and Objectives
The objectives of this project were to (i) evaluate HF patients for the presence of
malnutrition using four screening measures (i.e., albumin, prealbumin, Nutritional Risk
Index and Malnutrition Universal Screening Tool), and (ii) determine laboratory and co-
morbidity trends among malnourished patients. Based on these objectives, the primary
goal of this project was to test the performance of albumin, NRI and MUST in
comparison to prealbumin, among HF patients admitted to the University of Kentucky
Chandler Medical Center.
Methods
Human subject and research approval procedures
Once the project proposal was developed an expedited proposal was then
submitted and approved by the hospital’s Institutional Review Board (IRB). Noting the
project was a retrospective EMR review, patient consent was waived in compliance with
IRB regulations.
70
Study Setting
The study was conducted at the University of Kentucky Chandler Medical Center,
which is an 875-bed level 1 trauma center located in central Kentucky.
Study Design and Selection of Participants
A retrospective EMR review was performed. Inclusion criteria encompassed all
HF patients with the 428 diagnostic code, admitted from January 1, - December 31, 2013,
ages 18 and older, with all laboratory values available specified in the data collection tool
(Appendix A). The HF core measures coordinator provided a list of HF patients with
medical record numbers in order to obtain 100 patients who met the inclusion criteria. A
convenience sample was obtained of the first 100 patients from every HF diagnostic code
(428.0, 428.2, 428.3, and 428.4) who had all laboratory values available in their EMR.
The master list consisted of the medical record numbers for the 100 patients, who were
randomly assigned a study number.
In order to collect specific data via the EMR, a total of 100 patient records were
reviewed using the master list of medical record numbers. All data were collected
through the University of Kentucky’s secure network through the EMR program, Sunrise
Clinical Manager, which requires an active username and password to access. The
collected data included the following: demographics (gender, age, and ethnicity),
admitted unit (intensive care, telemetry, or progressive), HF diagnostic code, percent
ejection fraction (% EF), anthropometric measures (height, weight, BMI), presence of
unexplained weight loss, presence of acute illness or no nutritional intake >5 days,
dietitian consult, day of first dietitian note, intensive care unit (ICU) length of stay
(LOS), hospital LOS, diet order on admission, and dietary intake. Laboratory measures,
71
which included B-type natriuretic peptide (BNP), glomerular filtration rate (GFR),
glucose, HgbA1c, creatinine, albumin, and prealbumin, were obtained from the initial set
of labs acquired within the first 24 hours of admission. If the laboratory value wasn’t
available within the first 24 hours, the first available value was then used. Height,
weight, BMI and presence of recent weight loss were obtained from the adult patient
profile, which every patient must have completed within 24 hours of admission. Ejection
fraction was collected from results of the first echocardiogram conducted on admission.
The orders and documents sections of the patient’s EMR provided admission orders
which indicated to which unit the patient was admitted, transfer orders if the patient was
moved throughout their hospital stay, diet order, if a dietitian was consulted and when the
first nutrition note was documented.
Outcome Measure
For the purpose of this study, malnutrition or non-nutritional risk will be
classified as follows for each screening tool: the non-nutritional risk group will have
albumin > 3.2g/dL, prealbumin 11mg/dL, NRI score of 82.1, and a MUST score of 0;
the nutritionally at risk group will have albumin 3.2g/dL, prealbumin 10.9mg/dL,
NRI 82, and a MUST score of 1.
Instruments
Nutritional risk based on serum prealbumin can be classified into four categories:
normal is 16.0 – 35.0mg/dL, increased risk is 11.0 – 15.9mg/dL, significant risk 5.0 –
10.9mg/dl, and poor prognosis with < 5.0mg/dL (Prealbumin in Nutritional Care
Consensus Group, 1995). When the serum albumin level is 3.2g/dL a patient is at an
increased risk of being malnourished (Beck & Rosenthal, 2002).
72
The NRI was developed by the Veterans’ Affairs Total Parenteral Nutrition
Cooperative Study Group to determine nutritional risk in the postsurgical patient
population (Al-Najjar & Clark, 2012). The NRI uses the patient’s serum albumin, and
the ratio of current body weight to ideal body weight to predict a patient’s malnutrition
status. The score is calculated as follows: 1.5 x serum albumin + 41.7 x current
weight/ideal body weight. A score of > 100 means there is no evidence of malnutrition,
97.5 – 100 indicates mild malnutrition, 83.5 – 97.5 means moderate malnutrition, and <
83.5 signifies severe malnutrition (Al-Najjar & Clark, 2012).
The MUST was originally developed by the Malnutrition Advisory Group for the
British Association of Parenteral and Enteral Nutrition (Elia, 2010). MUST is a five-step
tool that is easy to use and usually takes 3-5 minutes to complete. It evaluates BMI
score, recent weight loss, and acute disease, then assigns an overall numerical risk (Elia,
2010). A score of 0 = low risk, 1 = medium risk, and ≥ 2 = high risk. Based on the
MUST score appropriate management guidelines are provided. A score of 0 requires no
intervention. Patients with a score of 1 require close dietary intake monitoring to evaluate
for necessary supplements. A score of 2 or more requires immediate nutritional
evaluation by a dietitian.
Data Analysis
Data analysis was performed using SPSS ® v. 21.0 (SPSS Inc., Chicago, IL).
Descriptive statistics. Data on patient age, gender, ethnicity, anthropometric
measurements, blood biochemical parameters, ICU and hospital LOS were analyzed
using descriptive statistics. In order to determine trends among patient demographics and
73
blood biochemical measures among the malnourished groups, descriptive statistics were
also computed using SPSS.
Consistency analysis. Sensitivity, specificity, positive predictive value (PPV),
negative predictive value (NPV), and confidence intervals (95%) were conducted to
compare the performance of serum albumin, MUST and NRI in comparison to serum
prealbumin levels. In the following equations, a represents test positive true cases; b
represents test positive not cases; c represents test negative true cases; and d represents
test negative not cases:
Sensitivity (Sn) = a / (a+c) Specificity (Sp) = b / (b+d)
PPV = a / (a+b) NPV = d / (d+c)
95% CI (Sn) = Sn ± 1.96 [Sn(1-Sn)] / a + c
95% CI (Sp) = Sp ± 1.96 [Sp(1-Sp)] / b + d
Results
Characteristics of the Study Population
The demographics of the participants are presented in Table 1. The mean age was
62.8 11.5 years, with 53% males involved in the study. The majority of patients were
diagnosed with systolic HF (49%), followed by diastolic HF (26%), then unspecified
congestive HF (23%), and finally combined diastolic and systolic HF (2%). A normal
EF% of > 55 was seen in 32% of patients, while 68% had a decreased EF% (< 55).
Seventy one patients were admitted to a telemetry unit, while eight went to a progressive
floor, and 21 were admitted to the ICU. Nineteen patients admitted to telemetry or
progressive floors were transferred into the ICU at some point during their hospital
admission. For all patients, the mean hospital LOS was 16 days (± 31). For those who
74
were in the ICU, fifteen patients stayed for 1 – 5 days, twelve remained for 6 – 10 days,
and thirteen stayed for > 10 days.
Aspects of Blood and Biochemical Parameters
Table 2 outlines the range of blood biochemical measures for all patients in the
study. The mean glucose level was 147mg/dL, while the mean albumin and prealbumin
levels were 2.8g/dL and 11.1mg/dL respectively. One patient had a normal BNP level,
indicating no signs of fluid volume overload upon admission, and eight patients showed
very little signs. The rest of the study population all showed some signs of fluid volume
overload associated with decompensated HF upon admission; seven patients indicated
mild decompensated HF, seven moderate, with 77 showing severe signs of unstable HF
upon admission. Thirty five patients had normal kidney function on admission with
creatinine levels less than 1.2mg/dL, while 65 showed signs of renal insufficiency with
levels > 1.2mg/dL. The GFR, another marker for renal function, was normal for 32
patients (> 60%), and abnormal for 68 patients (< 60%). A three month average of blood
glucose levels, HgA1c, was reported for all patients and indicates a patient’s risk for
developing diabetes mellitus. Based on those HgbA1c levels, 42 patients had normal
glucose levels over the last three months, with 26 indicating they were at risk for diabetes
and 32 were diabetic.
Malnutrition Prevalence
Analysis showed a range of malnutrition prevalence based on each screening
measure. Table 3 shows the prevalence of malnutrition among HF patients, based on
each tool’s malnutrition classifications and the limits set by this study. Analysis of serum
albumin levels revealed 21 patients were not at nutritional risk and 79 were at risk.
75
Serum prealbumin levels suggested 15 patients had no nutritional risk, 34 had a low
nutritional risk, 46 were at moderate risk, and five at high risk. Analysis using MUST
indicated 47 patients were not at nutritional risk, while 15 were at a low risk and 38 at a
high risk. Results of the NRI analysis indicated six patients had no nutritional risk, two
had a low risk, three had a moderate nutritional risk, and 89 were at a high risk. Based on
the study cut off limits for nutritional risk albumin, prealbumin, MUST and NRI found
79, 85, 53 and 92 patients to be at risk, respectively.
Characteristics of Malnourished Patients
Tables 4 and 5 provide the trends for malnourished patients with respect to patient
characteristics, and laboratory and biochemical measures. The mean ages for
malnourished patients according to each method were MUST 64.5 years, NRI 63.3 years,
prealbumin 63.2 years, and albumin 63.5 years (Table 4). Of the malnourished patients
identified by MUST, 27 (50.9%) were female and 26 male. For NRI 40 (43.5%)
malnourished patients were female and 52 male. Prealbumin identified 39 (45.9%)
female and 46 male patients, while albumin found 35 (44.3%) female and 44 male
patients. Dietitian consults on admission for those identified as malnourished were
ordered for 32 patients recognized by MUST, 39 patients per NRI, 41 based on
prealbumin, and 36 identified by albumin. Some patients were seen by a dietitian,
regardless if a consult was placed for routine screening, hospital length of stay or ICU
admission. Based on each tool, the number of identified malnourished patients seen by a
dietitian were as follows: MUST 48 (90.6%), NRI 67 (72.8%), prealbumin 47 (81%), and
albumin 59 (74.7%).
76
When analyzing hospital LOS for each screening method, MUST showed that
malnourished patients remained in the hospital for a mean of 21.1 days, while NRI
suggested they stayed 16.4 days, prealbumin indicated a mean of 16.9 days, and albumin
indicated malnourished patients stayed for 16.3 days. The mean EF for those
malnourished based on MUST, NRI, prealbumin, and albumin were 35.1%, 36.6%,
38.2%, and 39.7% respectively. The mean creatinine values for malnourished patients
were 1.5mg/dL according to MUST and NRI, while prealbumin and albumin observed a
mean of 1.6mg/dL. Renal function based on GFR was reduced for 33 (62.3%) patients
according to MUST, 63 (68.5%) per NRI, 58 (68.2%) according to prealbumin, and
54(68.4%) patients with albumin.
Comparison of Nutrition Screening Methods
Tables 6 shows the sensitivity, specificity, PPV and NPV values for each tool
compared to prealbumin, and Table 7 compares MUST to NRI. Sensitivity of a screening
tool suggests sensitiveness to a certain factor (Lalkhen & McCluskey, 2008). In this
study, test sensitivity was the proportion of at nutritional risk cases as diagnosed by
albumin, prealbumin, MUST and NRI. Specificity identifies the patients who are not at
nutritional risk and are classified appropriately (Lalkhen & McCluskey, 2008). A high
sensitivity may provide more false positives, or patients falsely identified as
malnourished, while a high specificity may give many false negatives. This means more
patients who are not malnourished may be classified as malnourished and may be subject
to extra treatment or testing. Conversely with a lower sensitivity and higher specificity,
malnourished patients may be misclassified as not malnourished and thus will not receive
77
appropriate treatment. For the purpose of this study, a higher sensitivity and a lower
specificity was desired.
MUST, NRI and albumin compared to prealbumin. In comparison to
prealbumin as a screening parameter, MUST revealed 49 true positive cases, four false
positives, 11 true negatives and 36 false negatives. There were 79 true positives, 13 false
positives, two true negatives and six false negatives with NRI. Serum albumin levels
showed 72 true positives, seven false positives, eight true negatives, and 13 false
negatives when compared to prealbumin as a screening parameter. The sensitivity of
MUST was 57.6% with a specificity of 73.3%, PPV 92.5%, and NPV of 23.4%. For
NRI, the sensitivity, specificity, PPV and NPV were 92.9%, 13.3%, 85.7%, and 25.0%
respectively. The sensitivity of albumin was 85.0%, with a specificity of 53.3%, PPV of
91.1% and NPV of 38.1%.
MUST compared to NRI. When comparing the two subjective screening tools
with NRI as the reference method, MUST had 50 true positives, three false positives, five
true negatives, and 42 false negatives. The sensitivity, specificity, PPV and NPV of
MUST were 54.3%, 62.5%, 94.3%, and 10.6% respectively.
Discussion
This project was designed to compare nutritional screening measures and evaluate
trends among malnourished HF patients. The results showed malnutrition prevalence to
be 53 - 92% based on the four screening tools. The prevalence among albumin,
prealbumin and NRI were similar, but MUST identified the fewest patients as
malnourished at 53%. The high incidence of malnutrition is not consistent with the
prevalence of 66% seen in the study conducted by Aziz et al. (2011). This may be
78
attributed to an increased prevalence of inflammation and infection seen in this
population, which was not evaluated in this study.
Malnutrition prevalence in male HF patients ranged from 49 - 57%, which is
pretty similar to the prevalence of 66% seen in other studies (Aziz et al., 2011). The
average LOS for malnourished patients in this study ranged from 16 – 22 days, which
appears to be much higher than those seen in other studies. For example, Aziz et al.
(2011) found the average hospital LOS to be 7 – 11 days for malnourished. The larger
range seen in this study may be attributed to a few outlying patients who had extremely
long lengths of stay ranging from 70 – 278 days. For patients who were in the ICU, 25 –
39% remained for 10 days, while 11 – 23% stayed for > 10 days. There were no
studies which measured ICU LOS, but rather focused solely on hospital LOS.
In terms of renal function, in this study 62 – 68% had a decreased GFR rate, and
elevated creatinine values averaging 1.5 – 1.6mg/dL. These measures of renal function
indicate that most of the malnourished patients experienced some sort of renal
dysfunction in addition to their HF. The average glucose values ranged from 134 –
156mg/dL, while the mean HgbA1c levels were 6.0 – 6.5%. These elevated glucose and
HgbA1c levels indicated most of the population was diabetic or at risk for becoming
diabetic. Upon admission BNP levels were collected for patients included in this study.
Based on those values 91% of the entire study population showed mild to severe
decompensated HF associated with fluid volume overload on admission to the hospital
(Table 2). Decompensated HF can worsen a patient’s prognosis and outcomes, and lead
to more hospital readmissions and cardiac cachexia (Araujo et al., 2011). The presence
of other comorbidities in addition to malnutrition and HF may also lead to worse
79
outcomes and a poor prognosis for these HF patients. Other studies that measured
specific laboratory values did not mention trends among the malnourished, so there is
little data available to determine patterns among the malnourished HF patient population.
Once patients are identified as malnourished or at risk for malnutrition, the next
step is providing appropriate treatment in order to correct the condition. Part of that step
is taking a multidisciplinary approach and involving a dietitian in the patient care plan.
Based on the results of this study, only 45 dietitian consults were placed at admission for
all the patients, but 74 patients were seen at some point during their admission by a
dietitian. When looking at patients identified as malnourished by the screening tools, 42
– 60% received a consult on admission and 73 – 91% were actually evaluated by a
dietitian. The low number of consults on admission for the malnourished patients is
concerning because even though a majority of those patients were eventually seen by an
RD, that first nutritional evaluation may have been delayed by a few days or even a week.
This delay in evaluation may lead to worsening malnutrition, a poor prognosis, and
increased morbidity and mortality.
In comparison to prealbumin, MUST found 36 false negatives, while NRI and
albumin only found six and thirteen respectively. When compared to NRI, MUST found
42 false negatives. This is concerning because 36 – 42 patients were not accurately
identified by MUST when they truly were malnourished.
Ideally a perfect screening tool would have a sensitivity of 100%, but this is
unrealistic. There is not a specific cut off for an adequate sensitivity range, but in general
≥ 85% sensitivity is acceptable in most of the literature (Aziz et al., 2011; Lomivorotov et
al., 2012; Van Venrooij et al., 2011). For this study, in striving for a higher sensitivity, a
80
lower specificity was acceptable in order to correctly identify the population as
malnourished. When compared to prealbumin the sensitivity of MUST was 57.6%, while
NRI and albumin were 92.9% and 84.7% respectively. NRI and albumin showed optimal
sensitivities compared to prealbumin in identifying patients as malnourished, while
MUST did not have the most favorable sensitivity. The specificities for MUST, NRI and
albumin were lower at 73.3%, 13.3%, and 53.3% respectively, but given the higher
sensitivities of NRI and albumin, these levels are more acceptable. These sensitivity and
specificity values can be attributed to the high incidence of true positives and low false
positives seen with NRI and albumin, and the moderate amount of true positives and false
negatives observed with MUST. Between the two subjective screening tools, NRI
performed best when compared to prealbumin given the high sensitivity level, even
though the specificity of MUST was higher than NRI. When comparing MUST to NRI,
the sensitivity and specificity remained less than optimal at 54.3% and 62.5%
respectively. This too can be attributed to the higher number of false negatives.
Overall the higher sensitivities of NRI (92.9% and 94.9%) mean it is the better
screening tool because there is a possibility that only 5-7% of patients who may be
malnourished were not correctly identified. The lower sensitivities of MUST at 57.6 %
and 54.3% indicate that there is a possibility that it misidentified 42-46% of patients as
not being malnourished. Currently no other research is available that compares these
tools to the laboratory markers of prealbumin and albumin in the HF patient population,
making it difficult to identify trends among the tools.
81
Limitations
As is the case with all studies, this study had a few limitations. First off, the chart
review was completed retrospectively, meaning all data collected are second hand
information. Height, weight and BMI may be inaccurate in that some measurements may
have been self-reported instead of accurately measured by the health provider. Presence
of recent weight loss, which is required for the MUST calculation, relied on the
admission patient profile being accurately completed by the patient’s nurse. The profile
information may have come from a family member of the patient who did not accurately
track the patient’s weight, or the patient may not recall recent weight loss over the past
six months. Both the BMI and recent weight loss inconsistencies could have affected the
overall MUST scores and their comparison to the other screening tools.
Another limitation is that albumin and prealbumin levels may not have been
collected immediately upon admission. Prealbumin has a half-life of two days, and
without adequate protein intake the value can decrease. These laboratory values may be
lower than normal in the presence of infection and inflammation. Markers for
inflammation and infection such as C-reactive protein and lymphocyte count were not
collected in this study, which may have been the reason for the increased prevalence of
malnutrition among HF patients.
Implications for Practice
Accurately identifying malnutrition in HF patients is difficult without a
standardized tool with which to evaluate patients. Some subjective tools and objective
measures work well in one population and not in others, such as HF patients. This study
has shown that NRI compares fairly well to prealbumin as a malnutrition screening tool
82
and that MUST was less optimal in terms of sensitivity and specificity. Implementing
NRI in combination with traditional laboratory screening measures could be beneficial in
identifying malnutrition in the HF population. Earlier identification of malnourished HF
patients on admission could lead to quicker nutrition evaluations by dietitians and
appropriate intervention. More rapid treatment of malnutrition could help improve
nutritional status among HF patients and may in turn help decrease hospital costs, LOS
and readmission rates.
Implications for Future Research
This project further identifies the need for a prospective study which evaluates a
large cohort of HF patients with a variety of subjective and objective screening measures.
Other studies may be helpful in narrowing down specific screening parameters which
work well in the HF patient population. If HF patients are accurately identified as being
malnourished, then other measures associated with malnutrition may be examined. Such
measures include outcomes, treatment options, laboratory and comorbidity trends among
the malnourished, and morbidity and mortality.
Conclusion
Heart failure is prevalent and associated with increased healthcare costs and
frequent hospital readmissions. Malnutrition is associated with a significant health risk
and financial burden. The development of malnutrition in the presence of HF will
worsen a patient’s myocardial dysfunction, decrease survival rates and lead to a poor
prognosis (Moughrabi & Evangelista, 2007). Early identification and treatment of
malnutrition in HF patients may help decrease associated healthcare costs and improve
outcomes. Quick, easy, inexpensive and reliable malnutrition screening methods may
83
help identify patients more quickly and accurately in order to reverse early malnutrition.
The results of this study provide evidence that NRI in combination with laboratory
measures may be helpful in identifying malnutrition among HF patients. There are a
number of nutritional screening tools available which are easy to use and inexpensive to
implement. Further research into NRI and other screening tools among the HF
population may be beneficial.
84
Tables
Table 1: Patient Demographics
Parameter Mean (SD)
Age (years) 62.8 (11.5)
Hospital LOS (days) 16 (31)
N
Gender
Male
Female
53
47
Ethnicity
Caucasian
African American
Other
77
18
5
BMI
≤ 18
19 – 24
25 – 29
30
3
30
20
47
HF Diagnostic Code
428.0 (unspecified CHF)
428.2 (systolic HF)
428.3 (diastolic HF)
428.4 (sys & dias HF)
23
49
26
2
Ejection Fraction (%)
< 55%
> 55%
68
32
Admit to:
Telemetry Bed
Progressive Bed
ICU
71
8
21
Hospital LOS
2 – 5 days
6 – 10 days
11 – 15 days
> 15 days
38
22
10
30
ICU LOS
0 days
1 – 5 days
6 – 10 days
> 10 days
60
15
12
13
Dietitian Consult
No
Yes
55
45
85
Table 1 (continued)
Parameter N
Dietitian Note
No
Yes
26
74
86
Table 2: Blood Biochemical Measures
Parameter Mean (SD)
Glucose (normal < 140mg/dL)
147 (83)
Albumin (normal 3.3-4.8g/dL)
2.8 (0.6)
Prealbumin (normal 16-35mg/dL)
11.1 (4.5)
N
BNP (pg/mL)
< 100 (no HF s/s)
100 – 300 (few HF s/s)
301 – 600 (mild HF)
601 – 900 (moderate HF s/s)
> 901 (severe HF s/s)
1
8
7
7
77
Creatinine (normal < 1.2mg/dL)
≤ 1.2
> 1.2
35
65
GFR (normal > 60%)
< 60
> 60
68
32
HgbA1c (%)
Normal (< 5.6)
At Risk (5.7 – 6.5)
Diabetic (> 6.5)
42
26
32
BNP – B-type Natriuretic Peptide; GFR – Glomerular Filtration Rate;
s/s – signs and symptoms
87
Table 3: Malnutrition Prevalence
Parameter N No Risk
Total (%)
At Risk
Total (%)
MUST Score
Low risk: 0
Mod risk: 1
High risk: ≥ 2
47
15
38
47 53
NRI score
No risk: > 98
Low risk: 92 – 98
Mod. risk: 82 – 91
High risk: < 82
6
2
3
89
8 92
Albumin
No risk: > 3.2g/dL
At risk: 3.2g/dL
21
79
21 79
Prealbumin
No risk: 16mg/dL
Low risk: 11 – 15.9mg/dL
Mod risk: 5 – 10.9mg/dL
High risk: < 5mg/dL
15
34
46
5
15 85
88
Table 4: Malnourished Patient Demographics
Study Characteristics % at risk: Number (%)
MUST
(n = 53)
NRI
(n = 92)
Prealbumin
(n = 85)
Albumin
(n = 79)
Gender
Female
Male
27 (50.9)
26 (49.1)
40 (43.5)
52 (56.5)
39 (45.9)
46 (54.1)
35 (44.3)
44 (55.7)
Race
Caucasian
African American
Other
44 (83.0)
6 (11.3)
3 (5.7)
71 (77.2)
16 (17.4)
5 (5.4)
66 (77.6)
15 (17.6)
4 (4.7)
63 (79.7)
12 (15.2)
4 (5.1)
Admit to:
Telemetry
Progressive
ICU
31 (58.5)
2 (3.8)
20 (37.7)
65 (70.7)
8 (8.7)
19 (20.7)
58 (68.2)
8 (9.4)
19 (22.4)
55 (69.6)
8 (10.1)
16 (20.3)
HF Diagnostic Code
428.0 (unspecified HF)
428.2 (systolic HF)
428.3 (diastolic HF)
428.4 (sys. & dia. HF)
10 (18.9)
27 (50.9)
14 (26.4)
2 (3.8)
19 (20.7)
48 (52.2)
23 (25.0)
2 (2.2)
23 (27.1)
40 (47.1)
20 (23.5)
2 (2.4)
22 (27.8)
34 (43.0)
21 (26.6)
2 (2.5)
ICU LOS
0 days
1 – 5 days
6 – 10 days
> 10 days
20 (37.7)
12 (22.7)
9 (17.0)
12 (22.7)
55 (59.8)
14 (15.2)
11 (12.1)
12 (13.2)
50 (58.8)
13 (15.3)
9 (10.7)
13 (15.3)
50 (63.3)
12 (15.2)
8 (10.1)
9 (11.4)
RD Consult
No
Yes
21 (39.6)
32 (60.4)
53 (57.6)
39 (42.4)
44 (51.8)
41 (48.2)
43 (54.4)
36 (45.6)
RD Note
No
Yes
5 (9.4)
48 (90.6)
25 (27.2)
67 (72.8)
11 (19.0)
47 (81.0)
20 (25.3)
59 (74.7)
89
Table 4 (continued)
Parameter % at risk: Mean (SD)
MUST
(n = 53)
NRI
(n = 92)
Prealbumin
(n = 85)
Albumin
(n = 79)
Age (years)
64.5 (12.2)
63.3 (11.6)
63.2 (11.6)
63.5 (11.5)
BMI
28.5 (10.9)
28.2 (6.8)
29.6 (8.9)
29.2 (8.9)
Hospital LOS (days)
22.1 (41.0)
16.4 (32.1)
16.9 (33.3)
16.3 (33.7)
EF %
35.1 (18.2)
36.6 (18.0)
38.2 (18.4)
39.7 (18.3)
90
Table 5: Malnourished Patient Lab and Biochemical Measures
Study Characteristics % at risk: Number (%)
MUST
(n = 53)
NRI
(n = 92)
Prealbumin
(n = 85)
Albumin
(n = 79)
GFR (%)
< 60
> 60
33 (62.3)
20 (37.7)
63 (68.5)
29 (31.5)
58 (68.2)
27 (31.8)
54 (68.4)
25 (31.6)
% at risk: Mean (SD)
MUST
(n = 53)
NRI
(n = 92)
Prealbumin
(n = 85)
Albumin
(n = 79)
Glucose (mg/dL)
133.9 (65.4)
144.9 (80.5)
149.1 (87.1)
152.9 (90.3)
BNP (pg/mL)
7665.0
(13003.7)
7921.1
(11576.9)
8557.9
(11892.2)
7872.4
(11652.6)
Creatinine (mg/dL)
1.5 (0.8)
1.5 (0.7)
1.6 (0.7)
1.6 (0.7)
HgbA1c (%)
6.0 (1.2)
6.3 (1.4)
6.4 (1.5)
6.5 (1.6)
Albumin (g/dL)
2.8 (0.5)
2.8 (0.6)
2.8 (0.6)
2.6 (0.5)
Prealbumin (mg/dL)
9.6 (3.9)
11.0 (4.4)
9.6 (3.0)
10.3 (4.0)
91
Table 6: Prediction Accuracy: Albumin, MUST and NRI compared to Prealbumin
Parameter Sensitivity
(CI) a
Specificity
(CI) a
PPV a
NPV a
Albumin (n = 100)
84.7
(77.1, 92.3)
53.3
(28.0, 78.6)
91.1
38.1
MUST (n = 100)
57.6
(46.5, 68.1)
73.3
(44.8, 91.1)
92.5
23.4
NRI (n = 100)
92.9
(84.7, 97.1)
13.3
(2.3, 41.6)
85.7
25.0
CI, Confidence Interval; PPV, Positive Predictive Value; NPV, Negative Predictive Value a
values are n (%)
Table 7: Prediction Accuracy: MUST compared to NRI
Parameter Sensitivity
(CI) a
Specificity
(CI) a
PPV a
NPV a
MUST (n = 100)
54.3
(43.7, 64.7)
62.5
(25.9, 89.9)
94.3
10.6
CI, Confidence Interval; PPV, Positive Predictive Value; NPV, Negative Predictive Value a
values are n (%)
92
References
Al-Najjar, Y. & Clark, A. L. (2012). Predicting outcome in patients with left ventricular
systolic chronic heart failure using a Nutritional Risk Index. American Journal of
Cardiology, 109, 1315-1320.
Araujo, J. P., Lourenco, P., Rocha-Gonocalves, F., Ferreira, A., & Bettencourt (2011).
Nutritional markers and prognosis in cardiac cachexia. International Journal of
Cardiology, 146, 359-363.
Aziz, E. F., Javed, F., Pratep, B., Musat, D., Nader, A., Pulimi, S., … Kukin, M. L.
(2011). Malnutrition as assessed by Nutritional Risk Index is associated with
worse outcome in patients admitted with acute decompensated heart failure: An
ACAP-HF data analysis. Heart International, 6(2), 3-8.
Beck, F. K., & Rosenthal, T. C. (2002). Prealbumin: A marker for nutritional evaluation.
American Family Physician, 65(8), 1575-1578.
Corkins, M. R., Guenter, P., DiMaria-Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S.,
… Resnick, H. E. (2014). Malnutrition diagnoses in hospitalized patients: United
States, 2010. Journal of Parenteral and Enteral Nutrition, 38, 186-195.
Elia, M. & Malnutrition Advisory Group (2010). The ‘MUST’ Report: Nutritional
screening of adults, a multidisciplinary responsibility. British Association of
Parenteral and Enteral Nutrition, 1-11.
Gout, B. S., Barker, L. A., & Crowe, T. C. (2009). Malnutrition identification, diagnosis
and dietetic referrals: Are we doing a good enough job? Nutrition and Dietetics,
66, 206-211.
Lalkhen, A. G., & McCluskey, A. (2008). Clinical tests: sensitivity and specificity.
Continuing Education in Anesthesia, Critical Care and Pain, 8(6), 221-223.
93
Le Moullac, B., Gouache, P., & Bleiberg-Daniel, F. (1992). Regulation of hepatic
transthyretin messenger RNA levels during moderate protein and food restriction
in rats. Journal of Nutrition, 122, 864-870.
Lomivorotov, V. V., Efremov, S. M., Boboshko, V. A., Nikolaev, D. A., Vedernikov, P.
E., Lomivorotov, V. N., & Karaskov, A. M. (2012). Evaluation of nutritional
screening tools for patients scheduled for cardiac surgery. Nutrition, 29, 436-442.
Moughrabi, S. M. & Evangelista, L. S. (2007). Cardiac cachexia at a glance. Progress in
Cardiovascular Nursing, Spring, 101-103.
Nicol, S. M., Carroll, D. L., Homeyer, C. M., & Zamagni, C. M. (2002). The
identification of malnutrition in heart failure patients. European Journal of
Cardiovascular Nursing, 1, 139-147.
Prealbumin in Nutritional Care Consensus Group (1995). Measurement of visceral
protein status in assessing protein and energy malnutrition: Standard of care.
Nutrition, 11, 169-171.
Sergi, G., Coin, A., Enzi, G., Volpato, S., Inelmen, E. M., Buttarello, M., … Bonometto,
P. (2006). Role of visceral proteins in detecting malnutrition in the elderly.
European Journal of Clinical Nutrition, 60, 203-209.
Van Venrooij, L. M., Van Leeuwen, P. A., Hopmans, W., Borgmeijer-Hoelen, M. M., De
Vos, R., & De Mol, B. A. (2011). Accuracy of quick and easy undernutrition
screening tools – Short Nutritional Assessment Questionnaire, Malnutrition
Universal Screening Tool, and Modified Malnutrition Universal Screening Tool –
in patients undergoing cardiac surgery. Journal of the American Dietetics
Association, 111, 1924-1930.
94
Conclusion to Final DNP Capstone Report
Cassondra Degener
University of Kentucky
95
Conclusion
This capstone evaluates HF patients admitted to the University of Kentucky
Chandler Medical Center for the presence of malnutrition. Four nutritional screening
measures were used to determine the best methods that can be used by health
professionals in order to accurately identify malnutrition in the HF population. This
study found malnutrition prevalence among HF patients to be fairly high at a rate of 53-
92%. The average hospital length of stay for malnourished patients was found to be 16-
22 days. Serum albumin, prealbumin and NRI demonstrated the best ability to screen HF
patients for malnutrition. Given these high rates of malnutrition, more needs to be done
in order to more accurately screen patients upon hospital admission and treat them
immediately. Early identification and treatment may help improve outcomes, decrease
hospital lengths of stay and overall healthcare costs.
Manuscript one concluded that based on the available literature, no one tool
demonstrated consistent reliability and validity in screening for malnutrition among
multiple patient populations in various healthcare settings. Malnutrition can occur in
many patient populations including chronic diseases such as cancer, liver failure and HF
(Jensen et al., 2010). The MUST demonstrated evidence of reliability and validity in
multiple patient populations such as cancer, chronic disease, the elderly, hospitalized
patients and outpatients. The use of MUST in the HF patient population is not well
described in the literature; however it has been used with success in other adult and
elderly populations. More research needs to be conducted within the HF population to
better identify a reliable and valid tool for this population.
96
Manuscript two revealed through a review of the literature that there is a lack of
studies in which multiple tools evaluated the nutritional status of HF patients. Of the two
studies reviewed that pertained to HF patients, one study compared NRI to traditional
nutritional biomarkers, while the second used NRI to evaluate HF patient outcomes.
Multiple subjective screening tools need to be studied within this population to better
identify malnutrition among HF patients. No one tool has been proven as the gold
standard of nutrition assessment, making it necessary to evaluate multiple tools in the HF
population.
Manuscript three showed that NRI compared well to prealbumin, while MUST
was less optimal in terms of sensitivity and specificity. Implementing NRI in
combination with traditional laboratory screening measures could be beneficial in
identifying malnutrition in the HF population. There are a number of nutritional
screening tools available which are easy to use and inexpensive to implement. Further
research into NRI and other screening tools among the HF population may be beneficial.
Overall this practice inquiry project has shown a high prevalence of malnutrition
in HF patients based on four nutritional screening methods. The need of future research
into effective screening tools in this population is necessary in order to accurately
identify malnutrition and take action to treat it as quickly as possible. Early intervention
may help increase quality of life and outcomes for HF patients.
97
Appendix
Appendix A: Data Collection Tool
Subject
ID #
Sex
(1=F
2=M)
Age
(yr)
Race
(1 = Caucasian,
2 = African
American,
3 = other)
Admit to:
(1 = ICU,
2 = tele,
3 = prog)
HF
diagnostic
code
%
EF
Height
(in)
Wt
(kg) BMI
Unexplained
wt loss
Pt acutely
ill or no
nutritional
intake >5
days
GFR Glu
cose
Hgb
A1c Creatinine
Albu
min BNP
Prealb
umin
RD
consult
(y/n)
RD note
(day #)
ICU
LOS
Hosp
LOS
Diet
Order
Diet
intake
Calculated
MUST
score
Calculated
NRI score
98
References
Araujo, J. P., Lourenco, P., Rocha-Gonocalves, F., Ferreira, A., & Bettencourt (2011).
Nutritional markers and prognosis in cardiac cachexia. International Journal of
Cardiology, 146, 359-363.
Arcand, J., Floras, V., Ahmed, M., Al-Hesayen, A., Ivanov, J., Allard, J. P., & Newton,
G. E. (2009). Nutritional inadequacies in patients with stable heart failure.
Journal of the American Dietetic Association, 109, 1909-1913.
Chaudhry, S. I., McAvay, G., Chen, S., Whitson, H., Newman, A. B., Krumholz, H. M.,
& Gill, T. M. (2012). Risk factors for hospitalization among older persons newly
diagnosed with heart failure: The cardiovascular health study. Journal of the
American College of Cardiology, 61(6), 635-642.
Corkins, M. R., Guenter, P., DiMaria-Ghalili, R. A., Jensen, G. L., Malone, A., Miller, S.,
… Resnick, H. E. (2014). Malnutrition diagnoses in hospitalized patients: United
States, 2010. Journal of Parenteral and Enteral Nutrition, 38, 186-195.
Go, A. S., Mozaffarian, D., Roger, V. L., Benjamin, E. J., Berry, J. D., Blaha, M. J. …
Turner, M. B. (2013). Heart disease and stroke statistics – 2014 update. A report
from the American Heart Association. Circulation, 129, e28-e292.
Jensen, G. L., Mirtallo, J., Compher, C., Dhaliwal, R., Forbese, A., Grijalba, R. F., …
Waitzberg, D. (2010). Adult starvation and disease-related malnutrition: A
proposal for etiology-based diagnosis in the clinical practice setting from the
International Consensus Guideline Committee. Journal of Parenteral and Enteral
Nutrition, 34(2), 156-159.
Kalantar-Zadeh, K., Anker, S. D., Horwich, T., B., & Fonarow, G. C. (2008). Nutritional
99
and anti-inflammatory in chronic heart failure. American Journal of Cardiology,
101, 89-103.
Lemon, S. C., Olendzki, B., Magner, R., Li, W., Culver, A. L., Ockene, I., & Goldberg,
R. J. (2009). The dietary quality of persons with heart failure in NHANES 1999-
2006. Journal of General Internal Medicine, 25(2), 135-140
Lennie, T. A., Moser, D. K., Heo, S., Chung, M. L., & Zambroski, C. H. (2006). Factors
influencing food intake in patients with heart failure. Journal of Cardiovascular
Nursing, 21(2), 123-129.
Machado d’Almeida, K. S., Perry, I. D., Clausell, N., & Souza, G. C. (2015). Adequacy
of energy and nutrient intake in patients with heart failure. Nutricion
Hospitalaria, 31, 500-507.
Nicol, S. M., Carroll, D. L., Homeyer, C. M., & Zamagni, C. M. (2002). The
identification of malnutrition in heart failure patients. European Journal of
Cardiovascular Nursing, 1, 139-147.